cuml.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Tour of cuml\n",
    "\n",
    "Machine learning is a fundamental capability of RAPIDS. cuML is a suite of libraries that implements a machine learning algorithms within the RAPIDS data science ecosystem. cuML enables data scientists, researchers, and software engineers to run traditional ML tasks on GPUs without going into the details of CUDA programming.\n",
    "\n",
    "The cuML repository contains:\n",
    "\n",
    "python: Python based GPU Dataframe (GDF) machine learning package that takes cuDF dataframes as input. cuML connects the data to C++/CUDA based cuML and ml-prims libraries without ever leaving GPU memory.\n",
    "\n",
    "cuML: C++/CUDA machine learning algorithms. This library currently includes the following six algorithms; a) Single GPU Truncated Singular Value Decomposition (tSVD), b) Single GPU Principal Component Analysis (PCA), c) Single GPU Density-based Spatial Clustering of Applications with Noise (DBSCAN), d) Single GPU Kalman Filtering, e) Multi-GPU K-Means Clustering, f) Multi-GPU K-Nearest Neighbors (Uses Faiss).\n",
    "\n",
    "ml-prims: Low level machine learning primitives used in cuML. ml-prims is comprised of the following components; a) Linear Algebra, b) Statistics, c) Basic Matrix Operations, d) Distance Functions, e) Random Number Generation.\n",
    "\n",
    "Available Algorithms:\n",
    "Truncated Singular Value Decomposition (tSVD).\n",
    "\n",
    "Principal Component Analysis (PCA).\n",
    "\n",
    "Density-based spatial clustering of applications with noise (DBSCAN).\n",
    "\n",
    "K-Means Clustering.\n",
    "\n",
    "K-Nearest Neighbors (Requires Faiss installation to use).\n",
    "\n",
    "Upcoming algorithms:\n",
    "\n",
    "Kalman Filter.\n",
    "More ML algorithms in cuML and more ML primitives in ml-prims are being added currently. Example notebooks are provided in the python folder to test the functionality and performance. Goals for future versions include more algorithms and multi-gpu versions of the algorithms and primitives."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Table of contents\n",
    "* Dimensionality reduction using PCA\n",
    "* Clustering\n",
    "    * k-means\n",
    "    * DBSCAN\n",
    "* Similarity searches\n",
    "    * k-nearest neighbors\n",
    "    * Truncated SVD"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "%matplotlib inline\n",
    "import os\n",
    "\n",
    "import cudf\n",
    "import cuml\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import sklearn.decomposition\n",
    "import sklearn.cluster\n",
    "import sklearn.neighbors"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Dimensionality reduction using PCA"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "home = os.path.expanduser('~')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style>\n",
       "    .dataframe thead tr:only-child th {\n",
       "        text-align: right;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: left;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>city</th>\n",
       "      <th>neighborhood</th>\n",
       "      <th>zipcode</th>\n",
       "      <th>price_per_m2</th>\n",
       "      <th>year_built</th>\n",
       "      <th>x</th>\n",
       "      <th>y</th>\n",
       "      <th>population</th>\n",
       "      <th>comprehensive_school</th>\n",
       "      <th>educated</th>\n",
       "      <th>...</th>\n",
       "      <th>small_houses</th>\n",
       "      <th>block_of_flats</th>\n",
       "      <th>labor_force</th>\n",
       "      <th>employed</th>\n",
       "      <th>unemployed</th>\n",
       "      <th>outside_labor_force</th>\n",
       "      <th>children</th>\n",
       "      <th>students</th>\n",
       "      <th>retired_persons</th>\n",
       "      <th>others</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>espoo</td>\n",
       "      <td>tapiola</td>\n",
       "      <td>2100</td>\n",
       "      <td>5444.022222</td>\n",
       "      <td>1985.111111</td>\n",
       "      <td>378711</td>\n",
       "      <td>6672696</td>\n",
       "      <td>4332</td>\n",
       "      <td>604</td>\n",
       "      <td>3003</td>\n",
       "      <td>...</td>\n",
       "      <td>299</td>\n",
       "      <td>2678</td>\n",
       "      <td>2083</td>\n",
       "      <td>1910</td>\n",
       "      <td>173</td>\n",
       "      <td>2085</td>\n",
       "      <td>485</td>\n",
       "      <td>257</td>\n",
       "      <td>1219</td>\n",
       "      <td>124</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>espoo</td>\n",
       "      <td>pohjois-tapiola</td>\n",
       "      <td>2130</td>\n",
       "      <td>3768.000000</td>\n",
       "      <td>1972.857143</td>\n",
       "      <td>377682</td>\n",
       "      <td>6674526</td>\n",
       "      <td>5983</td>\n",
       "      <td>651</td>\n",
       "      <td>3965</td>\n",
       "      <td>...</td>\n",
       "      <td>1414</td>\n",
       "      <td>1355</td>\n",
       "      <td>3033</td>\n",
       "      <td>2819</td>\n",
       "      <td>214</td>\n",
       "      <td>2895</td>\n",
       "      <td>1108</td>\n",
       "      <td>448</td>\n",
       "      <td>1172</td>\n",
       "      <td>167</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>espoo</td>\n",
       "      <td>laajalahti</td>\n",
       "      <td>2140</td>\n",
       "      <td>2770.000000</td>\n",
       "      <td>1977.750000</td>\n",
       "      <td>378239</td>\n",
       "      <td>6675892</td>\n",
       "      <td>3689</td>\n",
       "      <td>522</td>\n",
       "      <td>2190</td>\n",
       "      <td>...</td>\n",
       "      <td>1060</td>\n",
       "      <td>425</td>\n",
       "      <td>1900</td>\n",
       "      <td>1740</td>\n",
       "      <td>160</td>\n",
       "      <td>1810</td>\n",
       "      <td>848</td>\n",
       "      <td>311</td>\n",
       "      <td>550</td>\n",
       "      <td>101</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>espoo</td>\n",
       "      <td>westend</td>\n",
       "      <td>2160</td>\n",
       "      <td>4865.250000</td>\n",
       "      <td>1990.000000</td>\n",
       "      <td>378640</td>\n",
       "      <td>6670467</td>\n",
       "      <td>3194</td>\n",
       "      <td>286</td>\n",
       "      <td>2095</td>\n",
       "      <td>...</td>\n",
       "      <td>908</td>\n",
       "      <td>457</td>\n",
       "      <td>1442</td>\n",
       "      <td>1351</td>\n",
       "      <td>91</td>\n",
       "      <td>1713</td>\n",
       "      <td>622</td>\n",
       "      <td>270</td>\n",
       "      <td>654</td>\n",
       "      <td>167</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>espoo</td>\n",
       "      <td>haukilahti</td>\n",
       "      <td>2170</td>\n",
       "      <td>3860.937500</td>\n",
       "      <td>1972.937500</td>\n",
       "      <td>376534</td>\n",
       "      <td>6671246</td>\n",
       "      <td>5679</td>\n",
       "      <td>663</td>\n",
       "      <td>3865</td>\n",
       "      <td>...</td>\n",
       "      <td>1188</td>\n",
       "      <td>1678</td>\n",
       "      <td>2731</td>\n",
       "      <td>2510</td>\n",
       "      <td>221</td>\n",
       "      <td>2977</td>\n",
       "      <td>951</td>\n",
       "      <td>442</td>\n",
       "      <td>1390</td>\n",
       "      <td>194</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 37 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "    city     neighborhood  zipcode  price_per_m2   year_built       x  \\\n",
       "0  espoo          tapiola     2100   5444.022222  1985.111111  378711   \n",
       "1  espoo  pohjois-tapiola     2130   3768.000000  1972.857143  377682   \n",
       "2  espoo       laajalahti     2140   2770.000000  1977.750000  378239   \n",
       "3  espoo          westend     2160   4865.250000  1990.000000  378640   \n",
       "4  espoo       haukilahti     2170   3860.937500  1972.937500  376534   \n",
       "\n",
       "         y  population  comprehensive_school  educated   ...    small_houses  \\\n",
       "0  6672696        4332                   604      3003   ...             299   \n",
       "1  6674526        5983                   651      3965   ...            1414   \n",
       "2  6675892        3689                   522      2190   ...            1060   \n",
       "3  6670467        3194                   286      2095   ...             908   \n",
       "4  6671246        5679                   663      3865   ...            1188   \n",
       "\n",
       "   block_of_flats  labor_force  employed  unemployed  outside_labor_force  \\\n",
       "0            2678         2083      1910         173                 2085   \n",
       "1            1355         3033      2819         214                 2895   \n",
       "2             425         1900      1740         160                 1810   \n",
       "3             457         1442      1351          91                 1713   \n",
       "4            1678         2731      2510         221                 2977   \n",
       "\n",
       "   children  students  retired_persons  others  \n",
       "0       485       257             1219     124  \n",
       "1      1108       448             1172     167  \n",
       "2       848       311              550     101  \n",
       "3       622       270              654     167  \n",
       "4       951       442             1390     194  \n",
       "\n",
       "[5 rows x 37 columns]"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "path = os.path.join(home, 'demo/docker/data/full_apartments_data.csv')\n",
    "data = pd.read_csv(path)\n",
    "data.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "neighborhoods = data['neighborhood']\n",
    "data = data.drop(labels=['neighborhood', 'city', 'x', 'y'], axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "data = data.set_index('zipcode')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "32\n"
     ]
    }
   ],
   "source": [
    "print(len(data.columns))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "data = (data - data.mean()) / data.std()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "n_components = 2\n",
    "whiten = False\n",
    "random_state = 42\n",
    "svd_solver = \"full\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "pca_sk = sklearn.decomposition.PCA(n_components=n_components, svd_solver=svd_solver, \n",
    "                                   whiten=whiten, random_state=random_state)\n",
    "result_sk = pca_sk.fit_transform(data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0.61955611, 0.15456241])"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pca_sk.explained_variance_ratio_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "# GPU required.\n",
    "cuml_data = cudf.DataFrame.from_pandas(data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "           price_per_m2           year_built           population comprehensive_school            educated       matriculation          vocational ...              others\n",
      "2100   1.0904142660260845   0.5158729541840255  -0.8420522016760309  -0.7746147881478422 -0.7117857473760608 -0.5601331874585532 -0.9310043185910691 ... -0.8965710890770592\n",
      "2130 -0.12562506893884431 -0.07297100056264279 -0.46451880039781157  -0.7336140465761031 -0.3650112755934318 -0.4503020118587892 -0.8142813890960694 ... -0.7032887286878861\n",
      "2140  -0.8497248095054727  0.16214706815439525  -0.9890867122889291  -0.8461479968474722 -1.0048498279990103 -0.9400410243692123 -1.0952501550947473 ... -0.9999546771921982\n",
      "2160   0.6704855927212088   0.7508003350951634  -1.1022781318120778   -1.052024060909822 -1.0390947082686046 -0.8626190481267557 -1.2419875521741754 ... -0.7032887286878861\n",
      "2170 -0.05819418753712986 -0.06910957242947796  -0.5340343388524322  -0.7231457721322548 -0.4010585179824785 -0.4142917903506699 -0.7142331638146411 ... -0.5819253861179402\n",
      "2180 -0.12127176388533753   0.5906225753849996  -0.2808971642824814  -0.7065710042628284 -0.3433829301600038 -0.5745372760618009 -0.6900548427049625 ... -0.5504608158220283\n",
      "2200  -0.3599779909859595  -0.5226128546134371   0.6205181402836849 -0.19885969373618623  0.6122294655736236 0.17267482023167546 0.22705388904146365 ... 0.03837614257289417\n",
      "2210  -0.7078876815213936  0.24891002133374143   0.7252487870343962   0.2861703554954513  0.7881400084321714  0.5219739688604331  0.5872275000546056 ... 0.18670911682505023\n",
      "2230  0.06815142225537626   0.8613229892962851    2.831523868302523   2.1556296999260254  2.5807693724394625    2.01459765037198  2.3639172340106365 ...   2.766803881089825\n",
      "2260 -0.10240744198680814    1.039120302576356  -0.9845133216011251  -0.7807212815734205 -0.9951170725539678 -0.9202354025397467 -0.8426283862591407 ...  -0.869601457394849\n",
      "[83 more rows]\n",
      "[24 more columns]\n"
     ]
    }
   ],
   "source": [
    "print(cuml_data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Note: With this small dataset, the overhead of moving data to the GPU and launching the GPU kernels dominates\n",
    "# the run time and, thus, there is little benefit from using the GPU.\n",
    "pca_cuml = cuml.PCA(n_components=n_components, svd_solver=svd_solver, \n",
    "                    whiten=whiten, random_state=random_state)\n",
    "result_cuml = pca_cuml.fit_transform(cuml_data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "                    \n",
      "0 0.6195561070236225\n",
      "1 0.1545624063734113\n"
     ]
    }
   ],
   "source": [
    "# How good is the fit?\n",
    "print(pca_cuml.explained_variance_ratio_)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0, 0.5, 'Component 2')"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "result = result_cuml.to_pandas().values # CPU: use result_sk\n",
    "index = result_cuml.to_pandas().index # CPU: use data.index\n",
    "plt.figure(figsize=(20, 10))\n",
    "plt.plot(result[:, 0], result[:, 1], 'o')\n",
    "\n",
    "zipcodes = index.values\n",
    "\n",
    "for i in range(len(zipcodes)):\n",
    "    plt.text(result_sk[i, 0], result_sk[i, 1], neighborhoods[i])\n",
    "    \n",
    "plt.xlabel('Component 1')\n",
    "plt.ylabel('Component 2')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## k-means clustering"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<cudf.DataFrame ncols=2 nrows=93 >"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Note: We pipeline models, i.e, we take the GPU PCA result and feed it to the GPU k-means without going\n",
    "# through the CPU.\n",
    "result_cuml"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "labels:\n",
      "       \n",
      " 0    1\n",
      " 1    3\n",
      " 2    3\n",
      " 3    3\n",
      " 4    3\n",
      " 5    3\n",
      " 6    0\n",
      " 7    4\n",
      " 8    2\n",
      " 9    3\n",
      "[83 more rows]\n",
      "cluster_centers:\n",
      "                    0                     1\n",
      "0  0.2707176604069329 0.0028047242019252775\n",
      "1 -2.8474476819868517    -2.808827185648288\n",
      "2  10.890890497872057    2.3566885760786787\n",
      "3  -4.369022782584579    2.0134871080639596\n",
      "4   3.906592359684056   -0.8439355424936809\n"
     ]
    }
   ],
   "source": [
    "# GPU required.\n",
    "kmeans = cuml.KMeans(n_clusters=5, n_gpu=-1)\n",
    "kmeans.fit(result_cuml)\n",
    "print(\"labels:\")\n",
    "print(kmeans.labels_)\n",
    "print(\"cluster_centers:\")\n",
    "print(kmeans.cluster_centers_)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "# sklearn k-means may yield slightly different results due to differences in initialization and other parameters.\n",
    "#sk_kmeans = sklearn.cluster.KMeans(n_clusters=5)\n",
    "#sk_kmeans.fit(result_sk)\n",
    "#result = result_sk"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "colors = kmeans.labels_.to_pandas().values.astype('float')\n",
    "#colors = sk_kmeans.labels_.astype('float')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.collections.PathCollection at 0x7fce0c2a0898>"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.scatter(result[:, 0], result[:, 1], c=colors, cmap=\"jet\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Clustering using DBSCAN"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style>\n",
       "    .dataframe thead tr:only-child th {\n",
       "        text-align: right;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: left;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>game_id</th>\n",
       "      <th>player_id</th>\n",
       "      <th>team_id</th>\n",
       "      <th>timeOnIce</th>\n",
       "      <th>assists</th>\n",
       "      <th>goals</th>\n",
       "      <th>shots</th>\n",
       "      <th>hits</th>\n",
       "      <th>powerPlayGoals</th>\n",
       "      <th>powerPlayAssists</th>\n",
       "      <th>...</th>\n",
       "      <th>faceoffTaken</th>\n",
       "      <th>takeaways</th>\n",
       "      <th>giveaways</th>\n",
       "      <th>shortHandedGoals</th>\n",
       "      <th>shortHandedAssists</th>\n",
       "      <th>blocked</th>\n",
       "      <th>plusMinus</th>\n",
       "      <th>evenTimeOnIce</th>\n",
       "      <th>shortHandedTimeOnIce</th>\n",
       "      <th>powerPlayTimeOnIce</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2012030221</td>\n",
       "      <td>8471958</td>\n",
       "      <td>3</td>\n",
       "      <td>1925</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>5</td>\n",
       "      <td>0</td>\n",
       "      <td>1406</td>\n",
       "      <td>342</td>\n",
       "      <td>177</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2012030221</td>\n",
       "      <td>8471339</td>\n",
       "      <td>3</td>\n",
       "      <td>1597</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>1099</td>\n",
       "      <td>315</td>\n",
       "      <td>183</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2012030221</td>\n",
       "      <td>8471873</td>\n",
       "      <td>3</td>\n",
       "      <td>1695</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>1488</td>\n",
       "      <td>99</td>\n",
       "      <td>108</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2012030221</td>\n",
       "      <td>8473432</td>\n",
       "      <td>3</td>\n",
       "      <td>957</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>5</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>957</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2012030221</td>\n",
       "      <td>8470192</td>\n",
       "      <td>3</td>\n",
       "      <td>859</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>859</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 22 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "      game_id  player_id  team_id  timeOnIce  assists  goals  shots  hits  \\\n",
       "0  2012030221    8471958        3       1925        0      0      0     3   \n",
       "1  2012030221    8471339        3       1597        1      0      2     3   \n",
       "2  2012030221    8471873        3       1695        0      0      1     2   \n",
       "3  2012030221    8473432        3        957        0      0      3     5   \n",
       "4  2012030221    8470192        3        859        0      0      1     0   \n",
       "\n",
       "   powerPlayGoals  powerPlayAssists         ...          faceoffTaken  \\\n",
       "0               0                 0         ...                     0   \n",
       "1               0                 0         ...                     0   \n",
       "2               0                 0         ...                     0   \n",
       "3               0                 0         ...                     0   \n",
       "4               0                 0         ...                     0   \n",
       "\n",
       "   takeaways  giveaways  shortHandedGoals  shortHandedAssists  blocked  \\\n",
       "0          1          0                 0                   0        5   \n",
       "1          1          3                 0                   0        2   \n",
       "2          0          0                 0                   0        2   \n",
       "3          0          1                 0                   0        0   \n",
       "4          1          0                 0                   0        2   \n",
       "\n",
       "   plusMinus  evenTimeOnIce  shortHandedTimeOnIce  powerPlayTimeOnIce  \n",
       "0          0           1406                   342                 177  \n",
       "1          1           1099                   315                 183  \n",
       "2          0           1488                    99                 108  \n",
       "3          0            957                     0                   0  \n",
       "4          0            859                     0                   0  \n",
       "\n",
       "[5 rows x 22 columns]"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "path = os.path.join(home, 'demo/docker/data/game_skater_stats.csv')\n",
    "data = pd.read_csv(path)\n",
    "data.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "game_id                 0\n",
       "player_id               0\n",
       "team_id                 0\n",
       "timeOnIce               0\n",
       "assists                 0\n",
       "goals                   0\n",
       "shots                   0\n",
       "hits                    0\n",
       "powerPlayGoals          0\n",
       "powerPlayAssists        0\n",
       "penaltyMinutes          0\n",
       "faceOffWins             0\n",
       "faceoffTaken            0\n",
       "takeaways               0\n",
       "giveaways               0\n",
       "shortHandedGoals        0\n",
       "shortHandedAssists      0\n",
       "blocked                 0\n",
       "plusMinus               0\n",
       "evenTimeOnIce           0\n",
       "shortHandedTimeOnIce    0\n",
       "powerPlayTimeOnIce      0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data.isnull().sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "267854"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "data = data.drop(['game_id', 'team_id'], axis=1)\n",
    "data = data.set_index('player_id')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "data = data.astype('float')\n",
    "data = (data - data.mean()) / data.std()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "min_samples = 3\n",
    "eps = 3"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 0 ns, sys: 0 ns, total: 0 ns\n",
      "Wall time: 5.96 µs\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "# Not sure how long this takes... Enable at own risk or take a subset of the data.\n",
    "#clustering_sk = sklearn.cluster.DBSCAN(eps = eps, min_samples = min_samples, n_jobs=-1)\n",
    "#clustering_sk.fit(data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "#np.unique(clustering_sk.labels_)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "# GPU required.\n",
    "cuml_data = cudf.DataFrame.from_pandas(data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 31.7 s, sys: 3.2 s, total: 34.9 s\n",
      "Wall time: 34.6 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "# GPU required.\n",
    "clustering_cuml = cuml.DBSCAN(eps = eps, min_samples = min_samples)\n",
    "clustering_cuml.fit(cuml_data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "53"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# How many clusters?\n",
    "len(clustering_cuml.labels_.unique())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Put labels into the GPU df.\n",
    "cuml_data['cluster_label'] = clustering_cuml.labels_\n",
    "cuml_data['player_id'] = cuml_data.index.values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "             timeOnIce              assists              goals               shots                hits       powerPlayGoals     powerPlayAssists ... player_id\n",
      "8479339 -0.5348609619264693   1.4201681395403738 -0.376445486045435  0.2066818029181704 -0.8984266364520586 -0.17787727911885567 -0.24146235494694143 ...   8479339\n",
      "8479339  2.3333573780682424   1.4201681395403738 -0.376445486045435    3.46763484700443  -0.219577778279317 -0.17787727911885567    3.631623316168578 ...   8479339\n",
      "8479339 0.19157245831000352   1.4201681395403738 -0.376445486045435   4.119825455821682  -0.219577778279317 -0.17787727911885567 -0.24146235494694143 ...   8479339\n",
      "8479339  0.5360314979526503 -0.48933433694683975 2.1143511450304495  0.8588724117354224 -0.8984266364520586    5.266037337094805 -0.24146235494694143 ...   8479339\n",
      "8479339 -0.6883328112722029   1.4201681395403738 -0.376445486045435 -1.0976994147163335 0.45927107989342464 -0.17787727911885567 -0.24146235494694143 ...   8479339\n",
      "8479339  0.4541798449682589 -0.48933433694683975 -0.376445486045435  0.2066818029181704 -0.8984266364520586 -0.17787727911885567 -0.24146235494694143 ...   8479339\n",
      "8479339 -0.5485029040905344   1.4201681395403738 -0.376445486045435  0.8588724117354224 -0.8984266364520586 -0.17787727911885567    3.631623316168578 ...   8479339\n",
      "8479339 0.37914916306590024 -0.48933433694683975 2.1143511450304495  0.2066818029181704  2.4958176544116495    5.266037337094805 -0.24146235494694143 ...   8479339\n",
      "8479339  0.5496734401167155 -0.48933433694683975 -0.376445486045435  1.5110630205526743  1.1381199380661662 -0.17787727911885567 -0.24146235494694143 ...   8479339\n",
      "8479339  0.3484547931967535 -0.48933433694683975 -0.376445486045435  0.2066818029181704  1.1381199380661662 -0.17787727911885567 -0.24146235494694143 ...   8479339\n",
      "[162 more rows]\n",
      "[13 more columns]\n"
     ]
    }
   ],
   "source": [
    "# Let's see how Patrik Laine's games have been clustered.\n",
    "laine_id = 8479339 # Player ID.\n",
    "laine_data = cuml_data.query(\"player_id == @laine_id\")\n",
    "print(laine_data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "       \n",
      "8479339    0\n",
      "8479339    1\n",
      "8479339    0\n",
      "8479339    2\n",
      "8479339    0\n",
      "8479339    0\n",
      "8479339    1\n",
      "8479339    2\n",
      "8479339    0\n",
      "8479339    0\n",
      "[162 more rows]\n"
     ]
    }
   ],
   "source": [
    "print(laine_data['cluster_label'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "           timeOnIce              assists              goals             shots                hits     powerPlayGoals     powerPlayAssists ... player_id\n",
      "8479339  0.252961198048297 -0.48933433694683975 4.6051477761063335  3.46763484700443  -0.219577778279317 10.709951953308467 -0.24146235494694143 ...   8479339\n",
      "8479339 0.6451670352651719 -0.48933433694683975 4.6051477761063335 2.163253629369926  -0.219577778279317 10.709951953308467 -0.24146235494694143 ...   8479339\n",
      "8479339 0.5871887810678947 -0.48933433694683975 4.6051477761063335 4.119825455821682 -0.8984266364520586 10.709951953308467 -0.24146235494694143 ...   8479339\n",
      "[13 more columns]\n"
     ]
    }
   ],
   "source": [
    "# Most Patrik Laine's games are in the 0th cluster but there is some special cases like games with cluster 10.\n",
    "# Games with cluster 10 seem to contain a lot of goals.\n",
    "print(laine_data.query(\"cluster_label == 10\"))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Similarity searches"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## k-nearest neighbors"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The following examples are inspired by: https://beckernick.github.io/music_recommender/ and https://towardsdatascience.com/how-did-we-build-book-recommender-systems-in-an-hour-part-2-k-nearest-neighbors-and-matrix-c04b3c2ef55c"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [],
   "source": [
    "path = os.path.join(home, 'demo/docker/data/')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style>\n",
       "    .dataframe thead tr:only-child th {\n",
       "        text-align: right;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: left;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>users</th>\n",
       "      <th>artist-name</th>\n",
       "      <th>plays</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>00000c289a1829a808ac09c00daf10bc3c4e223b</td>\n",
       "      <td>betty blowtorch</td>\n",
       "      <td>2137</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>00000c289a1829a808ac09c00daf10bc3c4e223b</td>\n",
       "      <td>die Ärzte</td>\n",
       "      <td>1099</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>00000c289a1829a808ac09c00daf10bc3c4e223b</td>\n",
       "      <td>melissa etheridge</td>\n",
       "      <td>897</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>00000c289a1829a808ac09c00daf10bc3c4e223b</td>\n",
       "      <td>elvenking</td>\n",
       "      <td>717</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>00000c289a1829a808ac09c00daf10bc3c4e223b</td>\n",
       "      <td>juliette &amp; the licks</td>\n",
       "      <td>706</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                      users           artist-name  plays\n",
       "0  00000c289a1829a808ac09c00daf10bc3c4e223b       betty blowtorch   2137\n",
       "1  00000c289a1829a808ac09c00daf10bc3c4e223b             die Ärzte   1099\n",
       "2  00000c289a1829a808ac09c00daf10bc3c4e223b     melissa etheridge    897\n",
       "3  00000c289a1829a808ac09c00daf10bc3c4e223b             elvenking    717\n",
       "4  00000c289a1829a808ac09c00daf10bc3c4e223b  juliette & the licks    706"
      ]
     },
     "execution_count": 37,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# The initial data handling is copied from https://beckernick.github.io/music_recommender/, the rest of the\n",
    "# analysis is our own but inspired by the above blog posts.\n",
    "user_data = pd.read_table(os.path.join(path, 'usersha1-artmbid-artname-plays.tsv'),\n",
    "                          header=None, nrows=2e7,\n",
    "                          names=['users', 'musicbrainz-artist-id', 'artist-name', 'plays'],\n",
    "                          usecols=['users', 'artist-name', 'plays'])\n",
    "user_data.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style>\n",
       "    .dataframe thead tr:only-child th {\n",
       "        text-align: right;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: left;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>users</th>\n",
       "      <th>country</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>00000c289a1829a808ac09c00daf10bc3c4e223b</td>\n",
       "      <td>Germany</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>00001411dc427966b17297bf4d69e7e193135d89</td>\n",
       "      <td>Canada</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>00004d2ac9316e22dc007ab2243d6fcb239e707d</td>\n",
       "      <td>Germany</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>000063d3fe1cf2ba248b9e3c3f0334845a27a6bf</td>\n",
       "      <td>Mexico</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>00007a47085b9aab8af55f52ec8846ac479ac4fe</td>\n",
       "      <td>United States</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                      users        country\n",
       "0  00000c289a1829a808ac09c00daf10bc3c4e223b        Germany\n",
       "1  00001411dc427966b17297bf4d69e7e193135d89         Canada\n",
       "2  00004d2ac9316e22dc007ab2243d6fcb239e707d        Germany\n",
       "3  000063d3fe1cf2ba248b9e3c3f0334845a27a6bf         Mexico\n",
       "4  00007a47085b9aab8af55f52ec8846ac479ac4fe  United States"
      ]
     },
     "execution_count": 38,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "user_profiles = pd.read_table(os.path.join(path, 'usersha1-profile.tsv'),\n",
    "                              header=None,\n",
    "                              names=['users', 'gender', 'age', 'country', 'signup'],\n",
    "                              usecols=['users', 'country'])\n",
    "user_profiles.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [],
   "source": [
    "user_data = user_data.dropna(axis=0, subset=['artist-name'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style>\n",
       "    .dataframe thead tr:only-child th {\n",
       "        text-align: right;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: left;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>artist-name</th>\n",
       "      <th>total_artist_plays</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>04)]</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2</td>\n",
       "      <td>1606</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>58725ab=&gt;</td>\n",
       "      <td>23</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>80lİ yillarin tÜrkÇe sÖzlÜ aŞk Şarkilari</td>\n",
       "      <td>70</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>amy winehouse</td>\n",
       "      <td>23</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                 artist-name  total_artist_plays\n",
       "0                                       04)]                   6\n",
       "1                                         2                 1606\n",
       "2                                  58725ab=>                  23\n",
       "3   80lİ yillarin tÜrkÇe sÖzlÜ aŞk Şarkilari                  70\n",
       "4                              amy winehouse                  23"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "artist_plays = (user_data.\n",
    "     groupby(by = ['artist-name'])['plays'].\n",
    "     sum().\n",
    "     reset_index().\n",
    "     rename(columns = {'plays': 'total_artist_plays'})\n",
    "     [['artist-name', 'total_artist_plays']]\n",
    "    )\n",
    "artist_plays.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [],
   "source": [
    "user_data_with_artist_plays = user_data.merge(artist_plays,\n",
    "                                              left_on='artist-name',\n",
    "                                              right_on='artist-name',\n",
    "                                              how='left')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style>\n",
       "    .dataframe thead tr:only-child th {\n",
       "        text-align: right;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: left;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>users</th>\n",
       "      <th>artist-name</th>\n",
       "      <th>plays</th>\n",
       "      <th>total_artist_plays</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>00000c289a1829a808ac09c00daf10bc3c4e223b</td>\n",
       "      <td>betty blowtorch</td>\n",
       "      <td>2137</td>\n",
       "      <td>25651</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>00000c289a1829a808ac09c00daf10bc3c4e223b</td>\n",
       "      <td>die Ärzte</td>\n",
       "      <td>1099</td>\n",
       "      <td>3704875</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>00000c289a1829a808ac09c00daf10bc3c4e223b</td>\n",
       "      <td>melissa etheridge</td>\n",
       "      <td>897</td>\n",
       "      <td>180391</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>00000c289a1829a808ac09c00daf10bc3c4e223b</td>\n",
       "      <td>elvenking</td>\n",
       "      <td>717</td>\n",
       "      <td>410725</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>00000c289a1829a808ac09c00daf10bc3c4e223b</td>\n",
       "      <td>juliette &amp; the licks</td>\n",
       "      <td>706</td>\n",
       "      <td>90498</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                      users           artist-name  plays  \\\n",
       "0  00000c289a1829a808ac09c00daf10bc3c4e223b       betty blowtorch   2137   \n",
       "1  00000c289a1829a808ac09c00daf10bc3c4e223b             die Ärzte   1099   \n",
       "2  00000c289a1829a808ac09c00daf10bc3c4e223b     melissa etheridge    897   \n",
       "3  00000c289a1829a808ac09c00daf10bc3c4e223b             elvenking    717   \n",
       "4  00000c289a1829a808ac09c00daf10bc3c4e223b  juliette & the licks    706   \n",
       "\n",
       "   total_artist_plays  \n",
       "0               25651  \n",
       "1             3704875  \n",
       "2              180391  \n",
       "3              410725  \n",
       "4               90498  "
      ]
     },
     "execution_count": 42,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Note: The blog posts reduce the amount of data heavily due to computational constraints. Here, we use more\n",
    "# data as we can tap into the GPU.\n",
    "popularity_threshold = 5000\n",
    "user_data_popular_artists = user_data_with_artist_plays.query('total_artist_plays >= @popularity_threshold')\n",
    "user_data_popular_artists.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style>\n",
       "    .dataframe thead tr:only-child th {\n",
       "        text-align: right;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: left;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>users</th>\n",
       "      <th>artist-name</th>\n",
       "      <th>plays</th>\n",
       "      <th>total_artist_plays</th>\n",
       "      <th>country</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>277</th>\n",
       "      <td>0000ee7dd906373efa37f4e1185bfe1e3f8695ae</td>\n",
       "      <td>stam1na</td>\n",
       "      <td>758</td>\n",
       "      <td>791472</td>\n",
       "      <td>Finland</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>278</th>\n",
       "      <td>0000ee7dd906373efa37f4e1185bfe1e3f8695ae</td>\n",
       "      <td>dream theater</td>\n",
       "      <td>572</td>\n",
       "      <td>5892595</td>\n",
       "      <td>Finland</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>279</th>\n",
       "      <td>0000ee7dd906373efa37f4e1185bfe1e3f8695ae</td>\n",
       "      <td>ac/dc</td>\n",
       "      <td>554</td>\n",
       "      <td>6098979</td>\n",
       "      <td>Finland</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>280</th>\n",
       "      <td>0000ee7dd906373efa37f4e1185bfe1e3f8695ae</td>\n",
       "      <td>metallica</td>\n",
       "      <td>473</td>\n",
       "      <td>15481852</td>\n",
       "      <td>Finland</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>281</th>\n",
       "      <td>0000ee7dd906373efa37f4e1185bfe1e3f8695ae</td>\n",
       "      <td>iron maiden</td>\n",
       "      <td>471</td>\n",
       "      <td>9496576</td>\n",
       "      <td>Finland</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                        users    artist-name  plays  \\\n",
       "277  0000ee7dd906373efa37f4e1185bfe1e3f8695ae        stam1na    758   \n",
       "278  0000ee7dd906373efa37f4e1185bfe1e3f8695ae  dream theater    572   \n",
       "279  0000ee7dd906373efa37f4e1185bfe1e3f8695ae          ac/dc    554   \n",
       "280  0000ee7dd906373efa37f4e1185bfe1e3f8695ae      metallica    473   \n",
       "281  0000ee7dd906373efa37f4e1185bfe1e3f8695ae    iron maiden    471   \n",
       "\n",
       "     total_artist_plays  country  \n",
       "277              791472  Finland  \n",
       "278             5892595  Finland  \n",
       "279             6098979  Finland  \n",
       "280            15481852  Finland  \n",
       "281             9496576  Finland  "
      ]
     },
     "execution_count": 43,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "combined = user_data_popular_artists.merge(user_profiles,\n",
    "                                           left_on='users',\n",
    "                                           right_on='users',\n",
    "                                           how='left')\n",
    "finland_data = combined[combined[\"country\"] == \"Finland\"]\n",
    "finland_data.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [],
   "source": [
    "finland_data = finland_data.drop_duplicates(['users', 'artist-name'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 4.73 s, sys: 732 ms, total: 5.46 s\n",
      "Wall time: 4.13 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "X = finland_data.pivot_table(index='artist-name',\n",
    "                             columns='users',\n",
    "                             values='plays').fillna(0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Format the data to be suitable for cudf.\n",
    "X = X.reset_index()\n",
    "X = X.drop('artist-name', axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(20634, 11564)"
      ]
     },
     "execution_count": 47,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [],
   "source": [
    "X = X.astype(\"float32\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 39.6 s, sys: 3.23 s, total: 42.9 s\n",
      "Wall time: 43.2 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "# Expensive but CPU training below is even more expensive. GPU required.\n",
    "cuml_X = cudf.DataFrame.from_pandas(X)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [],
   "source": [
    "n_neighbors = 10"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 6.92 s, sys: 860 ms, total: 7.78 s\n",
      "Wall time: 7.78 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "# GPU required.\n",
    "cuml_knn = cuml.KNN(n_gpus=1)\n",
    "cuml_knn.fit(cuml_X)\n",
    "cuml_D, cuml_I = cuml_knn.query(cuml_X, n_neighbors)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 0 ns, sys: 0 ns, total: 0 ns\n",
      "Wall time: 3.58 µs\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "# Didn't finish in 1 hour on a 12-core CPU.\n",
    "# sk_knn = sklearn.neighbors.NearestNeighbors(n_neighbors).fit(X)\n",
    "# sk_D, sk_I = sk_knn.kneighbors(X, n_neighbors)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [],
   "source": [
    "#sk_I"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style>\n",
       "    .dataframe thead tr:only-child th {\n",
       "        text-align: right;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: left;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>index_neighbor_0</th>\n",
       "      <th>index_neighbor_1</th>\n",
       "      <th>index_neighbor_2</th>\n",
       "      <th>index_neighbor_3</th>\n",
       "      <th>index_neighbor_4</th>\n",
       "      <th>index_neighbor_5</th>\n",
       "      <th>index_neighbor_6</th>\n",
       "      <th>index_neighbor_7</th>\n",
       "      <th>index_neighbor_8</th>\n",
       "      <th>index_neighbor_9</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>19168</td>\n",
       "      <td>12052</td>\n",
       "      <td>2109</td>\n",
       "      <td>2844</td>\n",
       "      <td>19897</td>\n",
       "      <td>15370</td>\n",
       "      <td>112</td>\n",
       "      <td>2025</td>\n",
       "      <td>12451</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>11218</td>\n",
       "      <td>18651</td>\n",
       "      <td>5136</td>\n",
       "      <td>1962</td>\n",
       "      <td>13385</td>\n",
       "      <td>15076</td>\n",
       "      <td>1043</td>\n",
       "      <td>10887</td>\n",
       "      <td>9884</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>7948</td>\n",
       "      <td>19219</td>\n",
       "      <td>19086</td>\n",
       "      <td>11056</td>\n",
       "      <td>18627</td>\n",
       "      <td>17807</td>\n",
       "      <td>13977</td>\n",
       "      <td>12990</td>\n",
       "      <td>6093</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3</td>\n",
       "      <td>1043</td>\n",
       "      <td>1289</td>\n",
       "      <td>1419</td>\n",
       "      <td>3464</td>\n",
       "      <td>2698</td>\n",
       "      <td>3078</td>\n",
       "      <td>11138</td>\n",
       "      <td>9884</td>\n",
       "      <td>10887</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4</td>\n",
       "      <td>8966</td>\n",
       "      <td>2451</td>\n",
       "      <td>3371</td>\n",
       "      <td>4013</td>\n",
       "      <td>6363</td>\n",
       "      <td>11780</td>\n",
       "      <td>1483</td>\n",
       "      <td>16498</td>\n",
       "      <td>1383</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>5</td>\n",
       "      <td>3013</td>\n",
       "      <td>2979</td>\n",
       "      <td>11624</td>\n",
       "      <td>3093</td>\n",
       "      <td>14292</td>\n",
       "      <td>8728</td>\n",
       "      <td>2988</td>\n",
       "      <td>3229</td>\n",
       "      <td>12201</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>6</td>\n",
       "      <td>2786</td>\n",
       "      <td>11302</td>\n",
       "      <td>1289</td>\n",
       "      <td>3464</td>\n",
       "      <td>2698</td>\n",
       "      <td>3078</td>\n",
       "      <td>1043</td>\n",
       "      <td>9884</td>\n",
       "      <td>10887</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>7</td>\n",
       "      <td>6486</td>\n",
       "      <td>3043</td>\n",
       "      <td>12831</td>\n",
       "      <td>2797</td>\n",
       "      <td>5525</td>\n",
       "      <td>11417</td>\n",
       "      <td>18959</td>\n",
       "      <td>332</td>\n",
       "      <td>8841</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>8</td>\n",
       "      <td>17206</td>\n",
       "      <td>1289</td>\n",
       "      <td>1043</td>\n",
       "      <td>3464</td>\n",
       "      <td>2698</td>\n",
       "      <td>3078</td>\n",
       "      <td>1419</td>\n",
       "      <td>9884</td>\n",
       "      <td>10887</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>9</td>\n",
       "      <td>3672</td>\n",
       "      <td>2672</td>\n",
       "      <td>18865</td>\n",
       "      <td>3004</td>\n",
       "      <td>2698</td>\n",
       "      <td>3464</td>\n",
       "      <td>1043</td>\n",
       "      <td>9884</td>\n",
       "      <td>10887</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>10</td>\n",
       "      <td>6538</td>\n",
       "      <td>901</td>\n",
       "      <td>11625</td>\n",
       "      <td>16832</td>\n",
       "      <td>20107</td>\n",
       "      <td>10531</td>\n",
       "      <td>15463</td>\n",
       "      <td>6278</td>\n",
       "      <td>2599</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>11</td>\n",
       "      <td>13930</td>\n",
       "      <td>3023</td>\n",
       "      <td>10781</td>\n",
       "      <td>10516</td>\n",
       "      <td>694</td>\n",
       "      <td>670</td>\n",
       "      <td>6098</td>\n",
       "      <td>19556</td>\n",
       "      <td>9595</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>12</td>\n",
       "      <td>15864</td>\n",
       "      <td>10152</td>\n",
       "      <td>20334</td>\n",
       "      <td>6344</td>\n",
       "      <td>20370</td>\n",
       "      <td>20374</td>\n",
       "      <td>52</td>\n",
       "      <td>12696</td>\n",
       "      <td>20376</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>13</td>\n",
       "      <td>7738</td>\n",
       "      <td>15889</td>\n",
       "      <td>319</td>\n",
       "      <td>8059</td>\n",
       "      <td>18611</td>\n",
       "      <td>4912</td>\n",
       "      <td>6618</td>\n",
       "      <td>5567</td>\n",
       "      <td>4916</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>14</td>\n",
       "      <td>12404</td>\n",
       "      <td>9028</td>\n",
       "      <td>18428</td>\n",
       "      <td>18597</td>\n",
       "      <td>3459</td>\n",
       "      <td>2392</td>\n",
       "      <td>18128</td>\n",
       "      <td>12154</td>\n",
       "      <td>1939</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>15</td>\n",
       "      <td>18158</td>\n",
       "      <td>17769</td>\n",
       "      <td>8115</td>\n",
       "      <td>1547</td>\n",
       "      <td>18027</td>\n",
       "      <td>2378</td>\n",
       "      <td>19485</td>\n",
       "      <td>14641</td>\n",
       "      <td>1749</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>16</td>\n",
       "      <td>17146</td>\n",
       "      <td>8016</td>\n",
       "      <td>14917</td>\n",
       "      <td>2668</td>\n",
       "      <td>16155</td>\n",
       "      <td>17809</td>\n",
       "      <td>11032</td>\n",
       "      <td>7892</td>\n",
       "      <td>19154</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>17</td>\n",
       "      <td>9609</td>\n",
       "      <td>14152</td>\n",
       "      <td>1289</td>\n",
       "      <td>3464</td>\n",
       "      <td>2698</td>\n",
       "      <td>3078</td>\n",
       "      <td>1043</td>\n",
       "      <td>13067</td>\n",
       "      <td>14860</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>18</td>\n",
       "      <td>5145</td>\n",
       "      <td>16337</td>\n",
       "      <td>12045</td>\n",
       "      <td>4180</td>\n",
       "      <td>2698</td>\n",
       "      <td>3464</td>\n",
       "      <td>1289</td>\n",
       "      <td>10887</td>\n",
       "      <td>9884</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>19</td>\n",
       "      <td>2073</td>\n",
       "      <td>2383</td>\n",
       "      <td>6044</td>\n",
       "      <td>16530</td>\n",
       "      <td>19644</td>\n",
       "      <td>2507</td>\n",
       "      <td>1546</td>\n",
       "      <td>4339</td>\n",
       "      <td>12214</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20</th>\n",
       "      <td>20</td>\n",
       "      <td>291</td>\n",
       "      <td>8103</td>\n",
       "      <td>1289</td>\n",
       "      <td>3464</td>\n",
       "      <td>2698</td>\n",
       "      <td>1043</td>\n",
       "      <td>1419</td>\n",
       "      <td>9884</td>\n",
       "      <td>10887</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>21</td>\n",
       "      <td>1289</td>\n",
       "      <td>1043</td>\n",
       "      <td>1419</td>\n",
       "      <td>3464</td>\n",
       "      <td>2698</td>\n",
       "      <td>3078</td>\n",
       "      <td>11138</td>\n",
       "      <td>9884</td>\n",
       "      <td>10887</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22</th>\n",
       "      <td>22</td>\n",
       "      <td>10835</td>\n",
       "      <td>11155</td>\n",
       "      <td>8279</td>\n",
       "      <td>20192</td>\n",
       "      <td>3817</td>\n",
       "      <td>11987</td>\n",
       "      <td>19768</td>\n",
       "      <td>7202</td>\n",
       "      <td>11457</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23</th>\n",
       "      <td>23</td>\n",
       "      <td>800</td>\n",
       "      <td>6089</td>\n",
       "      <td>16109</td>\n",
       "      <td>11120</td>\n",
       "      <td>6638</td>\n",
       "      <td>5422</td>\n",
       "      <td>16875</td>\n",
       "      <td>15840</td>\n",
       "      <td>2175</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24</th>\n",
       "      <td>24</td>\n",
       "      <td>6108</td>\n",
       "      <td>16244</td>\n",
       "      <td>12009</td>\n",
       "      <td>19861</td>\n",
       "      <td>9173</td>\n",
       "      <td>11077</td>\n",
       "      <td>18859</td>\n",
       "      <td>775</td>\n",
       "      <td>9884</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25</th>\n",
       "      <td>25</td>\n",
       "      <td>11544</td>\n",
       "      <td>2234</td>\n",
       "      <td>547</td>\n",
       "      <td>3464</td>\n",
       "      <td>2698</td>\n",
       "      <td>1289</td>\n",
       "      <td>1043</td>\n",
       "      <td>10887</td>\n",
       "      <td>9884</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26</th>\n",
       "      <td>26</td>\n",
       "      <td>2634</td>\n",
       "      <td>18852</td>\n",
       "      <td>2595</td>\n",
       "      <td>19838</td>\n",
       "      <td>17857</td>\n",
       "      <td>2968</td>\n",
       "      <td>11126</td>\n",
       "      <td>19102</td>\n",
       "      <td>7106</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>27</th>\n",
       "      <td>27</td>\n",
       "      <td>1289</td>\n",
       "      <td>1043</td>\n",
       "      <td>1419</td>\n",
       "      <td>3464</td>\n",
       "      <td>2698</td>\n",
       "      <td>3078</td>\n",
       "      <td>11138</td>\n",
       "      <td>9884</td>\n",
       "      <td>10887</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>28</th>\n",
       "      <td>28</td>\n",
       "      <td>1043</td>\n",
       "      <td>1289</td>\n",
       "      <td>1419</td>\n",
       "      <td>3464</td>\n",
       "      <td>2698</td>\n",
       "      <td>3078</td>\n",
       "      <td>11138</td>\n",
       "      <td>9884</td>\n",
       "      <td>10887</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>29</th>\n",
       "      <td>29</td>\n",
       "      <td>8557</td>\n",
       "      <td>1043</td>\n",
       "      <td>1289</td>\n",
       "      <td>3464</td>\n",
       "      <td>2698</td>\n",
       "      <td>3078</td>\n",
       "      <td>1419</td>\n",
       "      <td>9884</td>\n",
       "      <td>10887</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20604</th>\n",
       "      <td>20604</td>\n",
       "      <td>20624</td>\n",
       "      <td>1419</td>\n",
       "      <td>1043</td>\n",
       "      <td>3464</td>\n",
       "      <td>2698</td>\n",
       "      <td>9884</td>\n",
       "      <td>1289</td>\n",
       "      <td>10887</td>\n",
       "      <td>14860</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20605</th>\n",
       "      <td>20605</td>\n",
       "      <td>5271</td>\n",
       "      <td>18506</td>\n",
       "      <td>17852</td>\n",
       "      <td>1318</td>\n",
       "      <td>2698</td>\n",
       "      <td>3464</td>\n",
       "      <td>1043</td>\n",
       "      <td>9884</td>\n",
       "      <td>10887</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20606</th>\n",
       "      <td>20606</td>\n",
       "      <td>6891</td>\n",
       "      <td>2798</td>\n",
       "      <td>20618</td>\n",
       "      <td>4714</td>\n",
       "      <td>20609</td>\n",
       "      <td>11967</td>\n",
       "      <td>4716</td>\n",
       "      <td>12734</td>\n",
       "      <td>10963</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20607</th>\n",
       "      <td>20607</td>\n",
       "      <td>14644</td>\n",
       "      <td>2201</td>\n",
       "      <td>11957</td>\n",
       "      <td>8970</td>\n",
       "      <td>19293</td>\n",
       "      <td>12109</td>\n",
       "      <td>20630</td>\n",
       "      <td>9788</td>\n",
       "      <td>2867</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20608</th>\n",
       "      <td>20608</td>\n",
       "      <td>20457</td>\n",
       "      <td>20593</td>\n",
       "      <td>1043</td>\n",
       "      <td>3464</td>\n",
       "      <td>2698</td>\n",
       "      <td>1419</td>\n",
       "      <td>1289</td>\n",
       "      <td>10887</td>\n",
       "      <td>14860</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20609</th>\n",
       "      <td>20609</td>\n",
       "      <td>20618</td>\n",
       "      <td>6891</td>\n",
       "      <td>11967</td>\n",
       "      <td>2798</td>\n",
       "      <td>8970</td>\n",
       "      <td>11957</td>\n",
       "      <td>19293</td>\n",
       "      <td>20630</td>\n",
       "      <td>2201</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20610</th>\n",
       "      <td>20610</td>\n",
       "      <td>2233</td>\n",
       "      <td>20616</td>\n",
       "      <td>14360</td>\n",
       "      <td>3464</td>\n",
       "      <td>2698</td>\n",
       "      <td>1419</td>\n",
       "      <td>1043</td>\n",
       "      <td>10887</td>\n",
       "      <td>13067</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20611</th>\n",
       "      <td>20611</td>\n",
       "      <td>1476</td>\n",
       "      <td>19789</td>\n",
       "      <td>20613</td>\n",
       "      <td>6565</td>\n",
       "      <td>2785</td>\n",
       "      <td>17698</td>\n",
       "      <td>15587</td>\n",
       "      <td>10887</td>\n",
       "      <td>14860</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20612</th>\n",
       "      <td>20612</td>\n",
       "      <td>17031</td>\n",
       "      <td>20591</td>\n",
       "      <td>15121</td>\n",
       "      <td>9822</td>\n",
       "      <td>17864</td>\n",
       "      <td>14383</td>\n",
       "      <td>20611</td>\n",
       "      <td>1476</td>\n",
       "      <td>19468</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20613</th>\n",
       "      <td>20613</td>\n",
       "      <td>19789</td>\n",
       "      <td>6565</td>\n",
       "      <td>2785</td>\n",
       "      <td>17698</td>\n",
       "      <td>15587</td>\n",
       "      <td>3464</td>\n",
       "      <td>1043</td>\n",
       "      <td>10887</td>\n",
       "      <td>14860</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20614</th>\n",
       "      <td>20614</td>\n",
       "      <td>20619</td>\n",
       "      <td>19199</td>\n",
       "      <td>20600</td>\n",
       "      <td>14028</td>\n",
       "      <td>7570</td>\n",
       "      <td>20475</td>\n",
       "      <td>20592</td>\n",
       "      <td>12531</td>\n",
       "      <td>19266</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20615</th>\n",
       "      <td>20615</td>\n",
       "      <td>16694</td>\n",
       "      <td>1419</td>\n",
       "      <td>1043</td>\n",
       "      <td>3464</td>\n",
       "      <td>2698</td>\n",
       "      <td>9884</td>\n",
       "      <td>1289</td>\n",
       "      <td>10887</td>\n",
       "      <td>14860</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20616</th>\n",
       "      <td>20616</td>\n",
       "      <td>20625</td>\n",
       "      <td>20608</td>\n",
       "      <td>1043</td>\n",
       "      <td>3464</td>\n",
       "      <td>1289</td>\n",
       "      <td>1419</td>\n",
       "      <td>3078</td>\n",
       "      <td>10887</td>\n",
       "      <td>14860</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20617</th>\n",
       "      <td>20617</td>\n",
       "      <td>1876</td>\n",
       "      <td>16978</td>\n",
       "      <td>1043</td>\n",
       "      <td>3464</td>\n",
       "      <td>2698</td>\n",
       "      <td>1419</td>\n",
       "      <td>1289</td>\n",
       "      <td>10887</td>\n",
       "      <td>14860</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20618</th>\n",
       "      <td>20618</td>\n",
       "      <td>6891</td>\n",
       "      <td>20609</td>\n",
       "      <td>11967</td>\n",
       "      <td>2798</td>\n",
       "      <td>20606</td>\n",
       "      <td>4716</td>\n",
       "      <td>1043</td>\n",
       "      <td>10887</td>\n",
       "      <td>14860</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20619</th>\n",
       "      <td>20619</td>\n",
       "      <td>19199</td>\n",
       "      <td>20600</td>\n",
       "      <td>14028</td>\n",
       "      <td>20475</td>\n",
       "      <td>20614</td>\n",
       "      <td>7570</td>\n",
       "      <td>20592</td>\n",
       "      <td>12531</td>\n",
       "      <td>19266</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20620</th>\n",
       "      <td>20620</td>\n",
       "      <td>2017</td>\n",
       "      <td>20628</td>\n",
       "      <td>20432</td>\n",
       "      <td>251</td>\n",
       "      <td>19410</td>\n",
       "      <td>1043</td>\n",
       "      <td>1289</td>\n",
       "      <td>10887</td>\n",
       "      <td>14860</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20621</th>\n",
       "      <td>20621</td>\n",
       "      <td>8842</td>\n",
       "      <td>7095</td>\n",
       "      <td>2609</td>\n",
       "      <td>20597</td>\n",
       "      <td>6149</td>\n",
       "      <td>20619</td>\n",
       "      <td>19199</td>\n",
       "      <td>1192</td>\n",
       "      <td>10055</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20622</th>\n",
       "      <td>20622</td>\n",
       "      <td>2201</td>\n",
       "      <td>11957</td>\n",
       "      <td>2867</td>\n",
       "      <td>11076</td>\n",
       "      <td>8970</td>\n",
       "      <td>19293</td>\n",
       "      <td>20630</td>\n",
       "      <td>20599</td>\n",
       "      <td>20609</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20623</th>\n",
       "      <td>20623</td>\n",
       "      <td>1043</td>\n",
       "      <td>1419</td>\n",
       "      <td>1289</td>\n",
       "      <td>3464</td>\n",
       "      <td>2698</td>\n",
       "      <td>9884</td>\n",
       "      <td>3078</td>\n",
       "      <td>10887</td>\n",
       "      <td>14860</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20624</th>\n",
       "      <td>20624</td>\n",
       "      <td>20604</td>\n",
       "      <td>16777</td>\n",
       "      <td>1043</td>\n",
       "      <td>3464</td>\n",
       "      <td>2698</td>\n",
       "      <td>1419</td>\n",
       "      <td>1289</td>\n",
       "      <td>10887</td>\n",
       "      <td>14860</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20625</th>\n",
       "      <td>20625</td>\n",
       "      <td>20608</td>\n",
       "      <td>1043</td>\n",
       "      <td>1289</td>\n",
       "      <td>3464</td>\n",
       "      <td>2698</td>\n",
       "      <td>1419</td>\n",
       "      <td>3078</td>\n",
       "      <td>10887</td>\n",
       "      <td>14860</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20626</th>\n",
       "      <td>20626</td>\n",
       "      <td>15272</td>\n",
       "      <td>10094</td>\n",
       "      <td>9167</td>\n",
       "      <td>6756</td>\n",
       "      <td>14722</td>\n",
       "      <td>5753</td>\n",
       "      <td>1419</td>\n",
       "      <td>10887</td>\n",
       "      <td>14860</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20627</th>\n",
       "      <td>20627</td>\n",
       "      <td>9199</td>\n",
       "      <td>1419</td>\n",
       "      <td>1043</td>\n",
       "      <td>3464</td>\n",
       "      <td>2698</td>\n",
       "      <td>9884</td>\n",
       "      <td>1289</td>\n",
       "      <td>10887</td>\n",
       "      <td>14860</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20628</th>\n",
       "      <td>20628</td>\n",
       "      <td>20432</td>\n",
       "      <td>251</td>\n",
       "      <td>19410</td>\n",
       "      <td>3464</td>\n",
       "      <td>1043</td>\n",
       "      <td>1419</td>\n",
       "      <td>1289</td>\n",
       "      <td>10887</td>\n",
       "      <td>14860</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20629</th>\n",
       "      <td>20629</td>\n",
       "      <td>20608</td>\n",
       "      <td>20593</td>\n",
       "      <td>20457</td>\n",
       "      <td>3464</td>\n",
       "      <td>2698</td>\n",
       "      <td>1043</td>\n",
       "      <td>1289</td>\n",
       "      <td>10887</td>\n",
       "      <td>14860</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20630</th>\n",
       "      <td>20630</td>\n",
       "      <td>19293</td>\n",
       "      <td>8970</td>\n",
       "      <td>11957</td>\n",
       "      <td>3464</td>\n",
       "      <td>2698</td>\n",
       "      <td>11138</td>\n",
       "      <td>1419</td>\n",
       "      <td>10887</td>\n",
       "      <td>9884</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20631</th>\n",
       "      <td>20631</td>\n",
       "      <td>20459</td>\n",
       "      <td>16068</td>\n",
       "      <td>13663</td>\n",
       "      <td>2186</td>\n",
       "      <td>20462</td>\n",
       "      <td>10072</td>\n",
       "      <td>1419</td>\n",
       "      <td>9884</td>\n",
       "      <td>10887</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20632</th>\n",
       "      <td>20632</td>\n",
       "      <td>1043</td>\n",
       "      <td>1419</td>\n",
       "      <td>1289</td>\n",
       "      <td>3464</td>\n",
       "      <td>2698</td>\n",
       "      <td>9884</td>\n",
       "      <td>3078</td>\n",
       "      <td>10887</td>\n",
       "      <td>14860</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20633</th>\n",
       "      <td>20633</td>\n",
       "      <td>2082</td>\n",
       "      <td>19069</td>\n",
       "      <td>2256</td>\n",
       "      <td>7346</td>\n",
       "      <td>2698</td>\n",
       "      <td>3464</td>\n",
       "      <td>1043</td>\n",
       "      <td>10887</td>\n",
       "      <td>14860</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>20634 rows × 10 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "       index_neighbor_0  index_neighbor_1  index_neighbor_2  index_neighbor_3  \\\n",
       "0                     0             19168             12052              2109   \n",
       "1                     1             11218             18651              5136   \n",
       "2                     2              7948             19219             19086   \n",
       "3                     3              1043              1289              1419   \n",
       "4                     4              8966              2451              3371   \n",
       "5                     5              3013              2979             11624   \n",
       "6                     6              2786             11302              1289   \n",
       "7                     7              6486              3043             12831   \n",
       "8                     8             17206              1289              1043   \n",
       "9                     9              3672              2672             18865   \n",
       "10                   10              6538               901             11625   \n",
       "11                   11             13930              3023             10781   \n",
       "12                   12             15864             10152             20334   \n",
       "13                   13              7738             15889               319   \n",
       "14                   14             12404              9028             18428   \n",
       "15                   15             18158             17769              8115   \n",
       "16                   16             17146              8016             14917   \n",
       "17                   17              9609             14152              1289   \n",
       "18                   18              5145             16337             12045   \n",
       "19                   19              2073              2383              6044   \n",
       "20                   20               291              8103              1289   \n",
       "21                   21              1289              1043              1419   \n",
       "22                   22             10835             11155              8279   \n",
       "23                   23               800              6089             16109   \n",
       "24                   24              6108             16244             12009   \n",
       "25                   25             11544              2234               547   \n",
       "26                   26              2634             18852              2595   \n",
       "27                   27              1289              1043              1419   \n",
       "28                   28              1043              1289              1419   \n",
       "29                   29              8557              1043              1289   \n",
       "...                 ...               ...               ...               ...   \n",
       "20604             20604             20624              1419              1043   \n",
       "20605             20605              5271             18506             17852   \n",
       "20606             20606              6891              2798             20618   \n",
       "20607             20607             14644              2201             11957   \n",
       "20608             20608             20457             20593              1043   \n",
       "20609             20609             20618              6891             11967   \n",
       "20610             20610              2233             20616             14360   \n",
       "20611             20611              1476             19789             20613   \n",
       "20612             20612             17031             20591             15121   \n",
       "20613             20613             19789              6565              2785   \n",
       "20614             20614             20619             19199             20600   \n",
       "20615             20615             16694              1419              1043   \n",
       "20616             20616             20625             20608              1043   \n",
       "20617             20617              1876             16978              1043   \n",
       "20618             20618              6891             20609             11967   \n",
       "20619             20619             19199             20600             14028   \n",
       "20620             20620              2017             20628             20432   \n",
       "20621             20621              8842              7095              2609   \n",
       "20622             20622              2201             11957              2867   \n",
       "20623             20623              1043              1419              1289   \n",
       "20624             20624             20604             16777              1043   \n",
       "20625             20625             20608              1043              1289   \n",
       "20626             20626             15272             10094              9167   \n",
       "20627             20627              9199              1419              1043   \n",
       "20628             20628             20432               251             19410   \n",
       "20629             20629             20608             20593             20457   \n",
       "20630             20630             19293              8970             11957   \n",
       "20631             20631             20459             16068             13663   \n",
       "20632             20632              1043              1419              1289   \n",
       "20633             20633              2082             19069              2256   \n",
       "\n",
       "       index_neighbor_4  index_neighbor_5  index_neighbor_6  index_neighbor_7  \\\n",
       "0                  2844             19897             15370               112   \n",
       "1                  1962             13385             15076              1043   \n",
       "2                 11056             18627             17807             13977   \n",
       "3                  3464              2698              3078             11138   \n",
       "4                  4013              6363             11780              1483   \n",
       "5                  3093             14292              8728              2988   \n",
       "6                  3464              2698              3078              1043   \n",
       "7                  2797              5525             11417             18959   \n",
       "8                  3464              2698              3078              1419   \n",
       "9                  3004              2698              3464              1043   \n",
       "10                16832             20107             10531             15463   \n",
       "11                10516               694               670              6098   \n",
       "12                 6344             20370             20374                52   \n",
       "13                 8059             18611              4912              6618   \n",
       "14                18597              3459              2392             18128   \n",
       "15                 1547             18027              2378             19485   \n",
       "16                 2668             16155             17809             11032   \n",
       "17                 3464              2698              3078              1043   \n",
       "18                 4180              2698              3464              1289   \n",
       "19                16530             19644              2507              1546   \n",
       "20                 3464              2698              1043              1419   \n",
       "21                 3464              2698              3078             11138   \n",
       "22                20192              3817             11987             19768   \n",
       "23                11120              6638              5422             16875   \n",
       "24                19861              9173             11077             18859   \n",
       "25                 3464              2698              1289              1043   \n",
       "26                19838             17857              2968             11126   \n",
       "27                 3464              2698              3078             11138   \n",
       "28                 3464              2698              3078             11138   \n",
       "29                 3464              2698              3078              1419   \n",
       "...                 ...               ...               ...               ...   \n",
       "20604              3464              2698              9884              1289   \n",
       "20605              1318              2698              3464              1043   \n",
       "20606              4714             20609             11967              4716   \n",
       "20607              8970             19293             12109             20630   \n",
       "20608              3464              2698              1419              1289   \n",
       "20609              2798              8970             11957             19293   \n",
       "20610              3464              2698              1419              1043   \n",
       "20611              6565              2785             17698             15587   \n",
       "20612              9822             17864             14383             20611   \n",
       "20613             17698             15587              3464              1043   \n",
       "20614             14028              7570             20475             20592   \n",
       "20615              3464              2698              9884              1289   \n",
       "20616              3464              1289              1419              3078   \n",
       "20617              3464              2698              1419              1289   \n",
       "20618              2798             20606              4716              1043   \n",
       "20619             20475             20614              7570             20592   \n",
       "20620               251             19410              1043              1289   \n",
       "20621             20597              6149             20619             19199   \n",
       "20622             11076              8970             19293             20630   \n",
       "20623              3464              2698              9884              3078   \n",
       "20624              3464              2698              1419              1289   \n",
       "20625              3464              2698              1419              3078   \n",
       "20626              6756             14722              5753              1419   \n",
       "20627              3464              2698              9884              1289   \n",
       "20628              3464              1043              1419              1289   \n",
       "20629              3464              2698              1043              1289   \n",
       "20630              3464              2698             11138              1419   \n",
       "20631              2186             20462             10072              1419   \n",
       "20632              3464              2698              9884              3078   \n",
       "20633              7346              2698              3464              1043   \n",
       "\n",
       "       index_neighbor_8  index_neighbor_9  \n",
       "0                  2025             12451  \n",
       "1                 10887              9884  \n",
       "2                 12990              6093  \n",
       "3                  9884             10887  \n",
       "4                 16498              1383  \n",
       "5                  3229             12201  \n",
       "6                  9884             10887  \n",
       "7                   332              8841  \n",
       "8                  9884             10887  \n",
       "9                  9884             10887  \n",
       "10                 6278              2599  \n",
       "11                19556              9595  \n",
       "12                12696             20376  \n",
       "13                 5567              4916  \n",
       "14                12154              1939  \n",
       "15                14641              1749  \n",
       "16                 7892             19154  \n",
       "17                13067             14860  \n",
       "18                10887              9884  \n",
       "19                 4339             12214  \n",
       "20                 9884             10887  \n",
       "21                 9884             10887  \n",
       "22                 7202             11457  \n",
       "23                15840              2175  \n",
       "24                  775              9884  \n",
       "25                10887              9884  \n",
       "26                19102              7106  \n",
       "27                 9884             10887  \n",
       "28                 9884             10887  \n",
       "29                 9884             10887  \n",
       "...                 ...               ...  \n",
       "20604             10887             14860  \n",
       "20605              9884             10887  \n",
       "20606             12734             10963  \n",
       "20607              9788              2867  \n",
       "20608             10887             14860  \n",
       "20609             20630              2201  \n",
       "20610             10887             13067  \n",
       "20611             10887             14860  \n",
       "20612              1476             19468  \n",
       "20613             10887             14860  \n",
       "20614             12531             19266  \n",
       "20615             10887             14860  \n",
       "20616             10887             14860  \n",
       "20617             10887             14860  \n",
       "20618             10887             14860  \n",
       "20619             12531             19266  \n",
       "20620             10887             14860  \n",
       "20621              1192             10055  \n",
       "20622             20599             20609  \n",
       "20623             10887             14860  \n",
       "20624             10887             14860  \n",
       "20625             10887             14860  \n",
       "20626             10887             14860  \n",
       "20627             10887             14860  \n",
       "20628             10887             14860  \n",
       "20629             10887             14860  \n",
       "20630             10887              9884  \n",
       "20631              9884             10887  \n",
       "20632             10887             14860  \n",
       "20633             10887             14860  \n",
       "\n",
       "[20634 rows x 10 columns]"
      ]
     },
     "execution_count": 54,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cuml_I.to_pandas()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Free GPU memory.\n",
    "for v in list(locals().keys()):\n",
    "    if 'cuml_' in v or '_cuml' in v:\n",
    "        del locals()[v]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Truncated singular value decomposition (TSVD)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [],
   "source": [
    "path = os.path.join(home, 'demo/docker/data/')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style>\n",
       "    .dataframe thead tr:only-child th {\n",
       "        text-align: right;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: left;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>users</th>\n",
       "      <th>artist-name</th>\n",
       "      <th>plays</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>00000c289a1829a808ac09c00daf10bc3c4e223b</td>\n",
       "      <td>betty blowtorch</td>\n",
       "      <td>2137</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>00000c289a1829a808ac09c00daf10bc3c4e223b</td>\n",
       "      <td>die Ärzte</td>\n",
       "      <td>1099</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>00000c289a1829a808ac09c00daf10bc3c4e223b</td>\n",
       "      <td>melissa etheridge</td>\n",
       "      <td>897</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>00000c289a1829a808ac09c00daf10bc3c4e223b</td>\n",
       "      <td>elvenking</td>\n",
       "      <td>717</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>00000c289a1829a808ac09c00daf10bc3c4e223b</td>\n",
       "      <td>juliette &amp; the licks</td>\n",
       "      <td>706</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                      users           artist-name  plays\n",
       "0  00000c289a1829a808ac09c00daf10bc3c4e223b       betty blowtorch   2137\n",
       "1  00000c289a1829a808ac09c00daf10bc3c4e223b             die Ärzte   1099\n",
       "2  00000c289a1829a808ac09c00daf10bc3c4e223b     melissa etheridge    897\n",
       "3  00000c289a1829a808ac09c00daf10bc3c4e223b             elvenking    717\n",
       "4  00000c289a1829a808ac09c00daf10bc3c4e223b  juliette & the licks    706"
      ]
     },
     "execution_count": 57,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "user_data = pd.read_table(os.path.join(path, 'usersha1-artmbid-artname-plays.tsv'),\n",
    "                          header=None, nrows=2e7,\n",
    "                          names=['users', 'musicbrainz-artist-id', 'artist-name', 'plays'],\n",
    "                          usecols=['users', 'artist-name', 'plays'])\n",
    "user_data.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style>\n",
       "    .dataframe thead tr:only-child th {\n",
       "        text-align: right;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: left;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>users</th>\n",
       "      <th>country</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>00000c289a1829a808ac09c00daf10bc3c4e223b</td>\n",
       "      <td>Germany</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>00001411dc427966b17297bf4d69e7e193135d89</td>\n",
       "      <td>Canada</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>00004d2ac9316e22dc007ab2243d6fcb239e707d</td>\n",
       "      <td>Germany</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>000063d3fe1cf2ba248b9e3c3f0334845a27a6bf</td>\n",
       "      <td>Mexico</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>00007a47085b9aab8af55f52ec8846ac479ac4fe</td>\n",
       "      <td>United States</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                      users        country\n",
       "0  00000c289a1829a808ac09c00daf10bc3c4e223b        Germany\n",
       "1  00001411dc427966b17297bf4d69e7e193135d89         Canada\n",
       "2  00004d2ac9316e22dc007ab2243d6fcb239e707d        Germany\n",
       "3  000063d3fe1cf2ba248b9e3c3f0334845a27a6bf         Mexico\n",
       "4  00007a47085b9aab8af55f52ec8846ac479ac4fe  United States"
      ]
     },
     "execution_count": 58,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "user_profiles = pd.read_table(os.path.join(path, 'usersha1-profile.tsv'),\n",
    "                              header=None,\n",
    "                              names=['users', 'gender', 'age', 'country', 'signup'],\n",
    "                              usecols=['users', 'country'])\n",
    "user_profiles.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [],
   "source": [
    "user_data = user_data.dropna(axis=0, subset=['artist-name'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style>\n",
       "    .dataframe thead tr:only-child th {\n",
       "        text-align: right;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: left;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>artist-name</th>\n",
       "      <th>total_artist_plays</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>04)]</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2</td>\n",
       "      <td>1606</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>58725ab=&gt;</td>\n",
       "      <td>23</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>80lİ yillarin tÜrkÇe sÖzlÜ aŞk Şarkilari</td>\n",
       "      <td>70</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>amy winehouse</td>\n",
       "      <td>23</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                 artist-name  total_artist_plays\n",
       "0                                       04)]                   6\n",
       "1                                         2                 1606\n",
       "2                                  58725ab=>                  23\n",
       "3   80lİ yillarin tÜrkÇe sÖzlÜ aŞk Şarkilari                  70\n",
       "4                              amy winehouse                  23"
      ]
     },
     "execution_count": 60,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "artist_plays = (user_data.\n",
    "     groupby(by = ['artist-name'])['plays'].\n",
    "     sum().\n",
    "     reset_index().\n",
    "     rename(columns = {'plays': 'total_artist_plays'})\n",
    "     [['artist-name', 'total_artist_plays']]\n",
    "    )\n",
    "artist_plays.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [],
   "source": [
    "user_data_with_artist_plays = user_data.merge(artist_plays,\n",
    "                                              left_on='artist-name',\n",
    "                                              right_on='artist-name',\n",
    "                                              how='left')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style>\n",
       "    .dataframe thead tr:only-child th {\n",
       "        text-align: right;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: left;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>users</th>\n",
       "      <th>artist-name</th>\n",
       "      <th>plays</th>\n",
       "      <th>total_artist_plays</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>00000c289a1829a808ac09c00daf10bc3c4e223b</td>\n",
       "      <td>die Ärzte</td>\n",
       "      <td>1099</td>\n",
       "      <td>3704875</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>00000c289a1829a808ac09c00daf10bc3c4e223b</td>\n",
       "      <td>red hot chili peppers</td>\n",
       "      <td>691</td>\n",
       "      <td>13547741</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>00000c289a1829a808ac09c00daf10bc3c4e223b</td>\n",
       "      <td>the black dahlia murder</td>\n",
       "      <td>507</td>\n",
       "      <td>1004743</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>00000c289a1829a808ac09c00daf10bc3c4e223b</td>\n",
       "      <td>walls of jericho</td>\n",
       "      <td>393</td>\n",
       "      <td>620899</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>00000c289a1829a808ac09c00daf10bc3c4e223b</td>\n",
       "      <td>goldfrapp</td>\n",
       "      <td>361</td>\n",
       "      <td>2577285</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                       users              artist-name  plays  \\\n",
       "1   00000c289a1829a808ac09c00daf10bc3c4e223b                die Ärzte   1099   \n",
       "5   00000c289a1829a808ac09c00daf10bc3c4e223b    red hot chili peppers    691   \n",
       "7   00000c289a1829a808ac09c00daf10bc3c4e223b  the black dahlia murder    507   \n",
       "10  00000c289a1829a808ac09c00daf10bc3c4e223b         walls of jericho    393   \n",
       "12  00000c289a1829a808ac09c00daf10bc3c4e223b                goldfrapp    361   \n",
       "\n",
       "    total_artist_plays  \n",
       "1              3704875  \n",
       "5             13547741  \n",
       "7              1004743  \n",
       "10              620899  \n",
       "12             2577285  "
      ]
     },
     "execution_count": 62,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "popularity_threshold = 500000\n",
    "user_data_popular_artists = user_data_with_artist_plays.query('total_artist_plays >= @popularity_threshold')\n",
    "user_data_popular_artists.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {},
   "outputs": [],
   "source": [
    "combined = user_data_popular_artists.merge(user_profiles,\n",
    "                                           left_on='users',\n",
    "                                           right_on='users',\n",
    "                                           how='left')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [],
   "source": [
    "combined = combined.drop_duplicates(['users', 'artist-name'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 18.3 s, sys: 2.88 s, total: 21.2 s\n",
      "Wall time: 18.2 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "X = combined.pivot_table(index='users', columns='artist-name', values='plays').fillna(0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style>\n",
       "    .dataframe thead tr:only-child th {\n",
       "        text-align: right;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: left;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th>artist-name</th>\n",
       "      <th>+44</th>\n",
       "      <th>...and you will know us by the trail of dead</th>\n",
       "      <th>16 horsepower</th>\n",
       "      <th>2pac</th>\n",
       "      <th>3 doors down</th>\n",
       "      <th>30 seconds to mars</th>\n",
       "      <th>311</th>\n",
       "      <th>36 crazyfists</th>\n",
       "      <th>3oh!3</th>\n",
       "      <th>50 cent</th>\n",
       "      <th>...</th>\n",
       "      <th>Сплин</th>\n",
       "      <th>ガゼット</th>\n",
       "      <th>ムック</th>\n",
       "      <th>久石譲</th>\n",
       "      <th>宇多田ヒカル</th>\n",
       "      <th>梶浦由記</th>\n",
       "      <th>植松伸夫</th>\n",
       "      <th>浜崎あゆみ</th>\n",
       "      <th>菅野よう子</th>\n",
       "      <th>동방신기</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>users</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>00000c289a1829a808ac09c00daf10bc3c4e223b</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>00001411dc427966b17297bf4d69e7e193135d89</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>00004d2ac9316e22dc007ab2243d6fcb239e707d</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>225.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>000063d3fe1cf2ba248b9e3c3f0334845a27a6bf</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>00007a47085b9aab8af55f52ec8846ac479ac4fe</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 1285 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "artist-name                               +44  \\\n",
       "users                                           \n",
       "00000c289a1829a808ac09c00daf10bc3c4e223b  0.0   \n",
       "00001411dc427966b17297bf4d69e7e193135d89  0.0   \n",
       "00004d2ac9316e22dc007ab2243d6fcb239e707d  0.0   \n",
       "000063d3fe1cf2ba248b9e3c3f0334845a27a6bf  0.0   \n",
       "00007a47085b9aab8af55f52ec8846ac479ac4fe  0.0   \n",
       "\n",
       "artist-name                               ...and you will know us by the trail of dead  \\\n",
       "users                                                                                    \n",
       "00000c289a1829a808ac09c00daf10bc3c4e223b                                           0.0   \n",
       "00001411dc427966b17297bf4d69e7e193135d89                                           0.0   \n",
       "00004d2ac9316e22dc007ab2243d6fcb239e707d                                           0.0   \n",
       "000063d3fe1cf2ba248b9e3c3f0334845a27a6bf                                           0.0   \n",
       "00007a47085b9aab8af55f52ec8846ac479ac4fe                                           0.0   \n",
       "\n",
       "artist-name                               16 horsepower  2pac  3 doors down  \\\n",
       "users                                                                         \n",
       "00000c289a1829a808ac09c00daf10bc3c4e223b            0.0   0.0           0.0   \n",
       "00001411dc427966b17297bf4d69e7e193135d89            0.0   0.0           0.0   \n",
       "00004d2ac9316e22dc007ab2243d6fcb239e707d          225.0   0.0           0.0   \n",
       "000063d3fe1cf2ba248b9e3c3f0334845a27a6bf            0.0   0.0           0.0   \n",
       "00007a47085b9aab8af55f52ec8846ac479ac4fe            0.0   0.0           0.0   \n",
       "\n",
       "artist-name                               30 seconds to mars  311  \\\n",
       "users                                                               \n",
       "00000c289a1829a808ac09c00daf10bc3c4e223b                 0.0  0.0   \n",
       "00001411dc427966b17297bf4d69e7e193135d89                 0.0  0.0   \n",
       "00004d2ac9316e22dc007ab2243d6fcb239e707d                 0.0  0.0   \n",
       "000063d3fe1cf2ba248b9e3c3f0334845a27a6bf                 0.0  0.0   \n",
       "00007a47085b9aab8af55f52ec8846ac479ac4fe                 0.0  0.0   \n",
       "\n",
       "artist-name                               36 crazyfists  3oh!3  50 cent  ...   \\\n",
       "users                                                                    ...    \n",
       "00000c289a1829a808ac09c00daf10bc3c4e223b            0.0    0.0      0.0  ...    \n",
       "00001411dc427966b17297bf4d69e7e193135d89            0.0    0.0      0.0  ...    \n",
       "00004d2ac9316e22dc007ab2243d6fcb239e707d            0.0    0.0      0.0  ...    \n",
       "000063d3fe1cf2ba248b9e3c3f0334845a27a6bf            0.0    0.0      0.0  ...    \n",
       "00007a47085b9aab8af55f52ec8846ac479ac4fe            0.0    0.0      0.0  ...    \n",
       "\n",
       "artist-name                               Сплин  ガゼット  ムック  久石譲  宇多田ヒカル  梶浦由記  \\\n",
       "users                                                                           \n",
       "00000c289a1829a808ac09c00daf10bc3c4e223b    0.0   0.0  0.0  0.0     0.0   0.0   \n",
       "00001411dc427966b17297bf4d69e7e193135d89    0.0   0.0  0.0  0.0     0.0   0.0   \n",
       "00004d2ac9316e22dc007ab2243d6fcb239e707d    0.0   0.0  0.0  0.0     0.0   0.0   \n",
       "000063d3fe1cf2ba248b9e3c3f0334845a27a6bf    0.0   0.0  0.0  0.0     0.0   0.0   \n",
       "00007a47085b9aab8af55f52ec8846ac479ac4fe    0.0   0.0  0.0  0.0     0.0   0.0   \n",
       "\n",
       "artist-name                               植松伸夫  浜崎あゆみ  菅野よう子  동방신기  \n",
       "users                                                               \n",
       "00000c289a1829a808ac09c00daf10bc3c4e223b   0.0    0.0    0.0   0.0  \n",
       "00001411dc427966b17297bf4d69e7e193135d89   0.0    0.0    0.0   0.0  \n",
       "00004d2ac9316e22dc007ab2243d6fcb239e707d   0.0    0.0    0.0   0.0  \n",
       "000063d3fe1cf2ba248b9e3c3f0334845a27a6bf   0.0    0.0    0.0   0.0  \n",
       "00007a47085b9aab8af55f52ec8846ac479ac4fe   0.0    0.0    0.0   0.0  \n",
       "\n",
       "[5 rows x 1285 columns]"
      ]
     },
     "execution_count": 66,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Format for cudf.\n",
    "X = X.reset_index()\n",
    "X = X.drop('users', axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(356741, 1285)"
      ]
     },
     "execution_count": 68,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "metadata": {},
   "outputs": [],
   "source": [
    "n_components = 40\n",
    "random_state = 42"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 2min 45s, sys: 1.92 s, total: 2min 46s\n",
      "Wall time: 28.3 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "algorithm = 'arpack'\n",
    "sk_tsvd = sklearn.decomposition.TruncatedSVD(n_components=n_components,\n",
    "                                             algorithm=algorithm, \n",
    "                                             random_state=random_state)\n",
    "sk_result = sk_tsvd.fit_transform(X)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 5.78 s, sys: 1.66 s, total: 7.44 s\n",
      "Wall time: 7.56 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "# The time to move data to the GPU + computing the TSVD is lower than that of computing the TSVD on the CPU.\n",
    "# GPU required.\n",
    "cuml_X = cudf.DataFrame.from_pandas(X)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 4.37 s, sys: 276 ms, total: 4.64 s\n",
      "Wall time: 4.64 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "# GPU required.\n",
    "algorithm = 'full'\n",
    "cuml_tsvd = cuml.TruncatedSVD(n_components=n_components,\n",
    "                              algorithm=algorithm, \n",
    "                              random_state=random_state)\n",
    "cuml_result = cuml_tsvd.fit_transform(cuml_X)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style>\n",
       "    .dataframe thead tr:only-child th {\n",
       "        text-align: right;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: left;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>0</th>\n",
       "      <th>1</th>\n",
       "      <th>2</th>\n",
       "      <th>3</th>\n",
       "      <th>4</th>\n",
       "      <th>5</th>\n",
       "      <th>6</th>\n",
       "      <th>7</th>\n",
       "      <th>8</th>\n",
       "      <th>9</th>\n",
       "      <th>...</th>\n",
       "      <th>30</th>\n",
       "      <th>31</th>\n",
       "      <th>32</th>\n",
       "      <th>33</th>\n",
       "      <th>34</th>\n",
       "      <th>35</th>\n",
       "      <th>36</th>\n",
       "      <th>37</th>\n",
       "      <th>38</th>\n",
       "      <th>39</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>25.484256</td>\n",
       "      <td>190.556608</td>\n",
       "      <td>-31.338465</td>\n",
       "      <td>-15.281125</td>\n",
       "      <td>-45.671815</td>\n",
       "      <td>-5.856145</td>\n",
       "      <td>6.885757</td>\n",
       "      <td>-67.207197</td>\n",
       "      <td>33.708774</td>\n",
       "      <td>19.864555</td>\n",
       "      <td>...</td>\n",
       "      <td>18.908646</td>\n",
       "      <td>2.617127</td>\n",
       "      <td>-25.639946</td>\n",
       "      <td>-140.907299</td>\n",
       "      <td>-138.664991</td>\n",
       "      <td>-101.156227</td>\n",
       "      <td>83.021828</td>\n",
       "      <td>-60.092995</td>\n",
       "      <td>174.518318</td>\n",
       "      <td>16.115740</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>136.531465</td>\n",
       "      <td>1683.418578</td>\n",
       "      <td>-245.665873</td>\n",
       "      <td>-706.399956</td>\n",
       "      <td>-113.498056</td>\n",
       "      <td>-15.900123</td>\n",
       "      <td>-17.033764</td>\n",
       "      <td>633.092502</td>\n",
       "      <td>-698.810449</td>\n",
       "      <td>1206.443617</td>\n",
       "      <td>...</td>\n",
       "      <td>119.943753</td>\n",
       "      <td>27.056856</td>\n",
       "      <td>1150.207759</td>\n",
       "      <td>402.881649</td>\n",
       "      <td>-240.484972</td>\n",
       "      <td>282.666337</td>\n",
       "      <td>-11.156897</td>\n",
       "      <td>-1598.418982</td>\n",
       "      <td>413.945688</td>\n",
       "      <td>-71.406816</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.847519</td>\n",
       "      <td>13.238973</td>\n",
       "      <td>0.447347</td>\n",
       "      <td>-3.803376</td>\n",
       "      <td>0.409061</td>\n",
       "      <td>-0.157565</td>\n",
       "      <td>-1.905522</td>\n",
       "      <td>0.135612</td>\n",
       "      <td>-8.502080</td>\n",
       "      <td>-4.277642</td>\n",
       "      <td>...</td>\n",
       "      <td>3.126801</td>\n",
       "      <td>2.241860</td>\n",
       "      <td>21.314967</td>\n",
       "      <td>-8.181559</td>\n",
       "      <td>16.993917</td>\n",
       "      <td>2.947127</td>\n",
       "      <td>9.914219</td>\n",
       "      <td>-4.323199</td>\n",
       "      <td>2.386330</td>\n",
       "      <td>2.616657</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.774074</td>\n",
       "      <td>8.552813</td>\n",
       "      <td>-2.787724</td>\n",
       "      <td>5.090695</td>\n",
       "      <td>-2.393326</td>\n",
       "      <td>-0.638501</td>\n",
       "      <td>0.677728</td>\n",
       "      <td>-10.803769</td>\n",
       "      <td>5.212503</td>\n",
       "      <td>-0.490907</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.239463</td>\n",
       "      <td>-0.025838</td>\n",
       "      <td>3.569100</td>\n",
       "      <td>0.399720</td>\n",
       "      <td>2.237289</td>\n",
       "      <td>1.984111</td>\n",
       "      <td>-3.765371</td>\n",
       "      <td>1.062166</td>\n",
       "      <td>-4.699627</td>\n",
       "      <td>9.414097</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>7.409898</td>\n",
       "      <td>103.713331</td>\n",
       "      <td>-15.019988</td>\n",
       "      <td>-47.338064</td>\n",
       "      <td>-8.188895</td>\n",
       "      <td>-0.580236</td>\n",
       "      <td>-12.552942</td>\n",
       "      <td>45.456363</td>\n",
       "      <td>-89.877750</td>\n",
       "      <td>57.267064</td>\n",
       "      <td>...</td>\n",
       "      <td>18.248195</td>\n",
       "      <td>11.618561</td>\n",
       "      <td>80.990458</td>\n",
       "      <td>-15.497598</td>\n",
       "      <td>21.382244</td>\n",
       "      <td>-5.109917</td>\n",
       "      <td>15.604298</td>\n",
       "      <td>-59.106464</td>\n",
       "      <td>29.079635</td>\n",
       "      <td>22.248695</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>105.730569</td>\n",
       "      <td>1557.161201</td>\n",
       "      <td>-305.962548</td>\n",
       "      <td>-736.052351</td>\n",
       "      <td>-218.394935</td>\n",
       "      <td>6.163846</td>\n",
       "      <td>-167.804834</td>\n",
       "      <td>852.814889</td>\n",
       "      <td>-960.758184</td>\n",
       "      <td>908.356908</td>\n",
       "      <td>...</td>\n",
       "      <td>-26.392069</td>\n",
       "      <td>-23.784510</td>\n",
       "      <td>-70.650686</td>\n",
       "      <td>-5.380971</td>\n",
       "      <td>147.839625</td>\n",
       "      <td>-216.426469</td>\n",
       "      <td>-82.874996</td>\n",
       "      <td>1.231659</td>\n",
       "      <td>-137.057685</td>\n",
       "      <td>140.238067</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>29.349910</td>\n",
       "      <td>274.144686</td>\n",
       "      <td>-83.792972</td>\n",
       "      <td>131.760430</td>\n",
       "      <td>-92.864841</td>\n",
       "      <td>-20.080335</td>\n",
       "      <td>29.475516</td>\n",
       "      <td>-379.396561</td>\n",
       "      <td>299.740749</td>\n",
       "      <td>70.651411</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.383982</td>\n",
       "      <td>2.354587</td>\n",
       "      <td>-2.417050</td>\n",
       "      <td>172.204439</td>\n",
       "      <td>2.921642</td>\n",
       "      <td>116.451911</td>\n",
       "      <td>-8.540848</td>\n",
       "      <td>-45.587263</td>\n",
       "      <td>-54.867679</td>\n",
       "      <td>26.263262</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>3.198312</td>\n",
       "      <td>36.748252</td>\n",
       "      <td>25.376836</td>\n",
       "      <td>9.704492</td>\n",
       "      <td>-8.451919</td>\n",
       "      <td>-1.218361</td>\n",
       "      <td>-2.712137</td>\n",
       "      <td>-31.168713</td>\n",
       "      <td>15.768606</td>\n",
       "      <td>-1.285726</td>\n",
       "      <td>...</td>\n",
       "      <td>0.380364</td>\n",
       "      <td>2.389711</td>\n",
       "      <td>11.452508</td>\n",
       "      <td>3.563717</td>\n",
       "      <td>0.498055</td>\n",
       "      <td>7.156311</td>\n",
       "      <td>-6.503403</td>\n",
       "      <td>-5.383621</td>\n",
       "      <td>-0.662745</td>\n",
       "      <td>-13.626061</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>0.067702</td>\n",
       "      <td>1.186290</td>\n",
       "      <td>1.295021</td>\n",
       "      <td>0.007708</td>\n",
       "      <td>-0.190129</td>\n",
       "      <td>-0.174863</td>\n",
       "      <td>1.653260</td>\n",
       "      <td>0.010459</td>\n",
       "      <td>0.038666</td>\n",
       "      <td>0.359853</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.121605</td>\n",
       "      <td>-0.178907</td>\n",
       "      <td>-0.324101</td>\n",
       "      <td>0.524534</td>\n",
       "      <td>-0.726809</td>\n",
       "      <td>-0.414440</td>\n",
       "      <td>-0.419759</td>\n",
       "      <td>-0.019525</td>\n",
       "      <td>-0.000276</td>\n",
       "      <td>0.198532</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>71.061732</td>\n",
       "      <td>256.103567</td>\n",
       "      <td>-48.734459</td>\n",
       "      <td>10.349713</td>\n",
       "      <td>-81.659392</td>\n",
       "      <td>-20.028944</td>\n",
       "      <td>53.286094</td>\n",
       "      <td>-194.765781</td>\n",
       "      <td>-1.964136</td>\n",
       "      <td>47.967224</td>\n",
       "      <td>...</td>\n",
       "      <td>-30.107370</td>\n",
       "      <td>-63.556273</td>\n",
       "      <td>-431.644919</td>\n",
       "      <td>51.788518</td>\n",
       "      <td>128.044258</td>\n",
       "      <td>38.745012</td>\n",
       "      <td>197.894843</td>\n",
       "      <td>111.665653</td>\n",
       "      <td>-266.072087</td>\n",
       "      <td>-94.960629</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>57.222351</td>\n",
       "      <td>673.416804</td>\n",
       "      <td>32.139426</td>\n",
       "      <td>-215.983043</td>\n",
       "      <td>-77.710093</td>\n",
       "      <td>-4.450906</td>\n",
       "      <td>-58.813590</td>\n",
       "      <td>164.617646</td>\n",
       "      <td>-287.319744</td>\n",
       "      <td>411.922621</td>\n",
       "      <td>...</td>\n",
       "      <td>34.912641</td>\n",
       "      <td>-16.006519</td>\n",
       "      <td>214.332887</td>\n",
       "      <td>-20.128511</td>\n",
       "      <td>131.563741</td>\n",
       "      <td>106.336729</td>\n",
       "      <td>97.776656</td>\n",
       "      <td>-232.446729</td>\n",
       "      <td>48.091937</td>\n",
       "      <td>-182.587256</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>9.393029</td>\n",
       "      <td>124.085309</td>\n",
       "      <td>17.094020</td>\n",
       "      <td>-48.434145</td>\n",
       "      <td>-15.518081</td>\n",
       "      <td>-2.077937</td>\n",
       "      <td>70.038885</td>\n",
       "      <td>90.884239</td>\n",
       "      <td>67.141310</td>\n",
       "      <td>-24.036977</td>\n",
       "      <td>...</td>\n",
       "      <td>14.876476</td>\n",
       "      <td>4.032670</td>\n",
       "      <td>225.302064</td>\n",
       "      <td>69.192388</td>\n",
       "      <td>31.292021</td>\n",
       "      <td>-51.494772</td>\n",
       "      <td>19.965973</td>\n",
       "      <td>114.762513</td>\n",
       "      <td>-70.685876</td>\n",
       "      <td>-0.977672</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>0.812545</td>\n",
       "      <td>8.432800</td>\n",
       "      <td>-0.739086</td>\n",
       "      <td>-1.982202</td>\n",
       "      <td>-1.787722</td>\n",
       "      <td>-0.252183</td>\n",
       "      <td>1.456411</td>\n",
       "      <td>-0.118746</td>\n",
       "      <td>-2.127982</td>\n",
       "      <td>1.341633</td>\n",
       "      <td>...</td>\n",
       "      <td>33.565733</td>\n",
       "      <td>-16.045833</td>\n",
       "      <td>0.424453</td>\n",
       "      <td>-0.633023</td>\n",
       "      <td>-0.423473</td>\n",
       "      <td>-0.708624</td>\n",
       "      <td>-0.833846</td>\n",
       "      <td>-1.892525</td>\n",
       "      <td>1.415557</td>\n",
       "      <td>2.027099</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>2.821769</td>\n",
       "      <td>40.690979</td>\n",
       "      <td>49.265773</td>\n",
       "      <td>6.736184</td>\n",
       "      <td>-7.020680</td>\n",
       "      <td>-5.993160</td>\n",
       "      <td>127.343407</td>\n",
       "      <td>4.896962</td>\n",
       "      <td>3.774986</td>\n",
       "      <td>9.026917</td>\n",
       "      <td>...</td>\n",
       "      <td>31.507425</td>\n",
       "      <td>56.899463</td>\n",
       "      <td>-8.361463</td>\n",
       "      <td>4.232159</td>\n",
       "      <td>3.130651</td>\n",
       "      <td>5.321420</td>\n",
       "      <td>2.812752</td>\n",
       "      <td>-0.126850</td>\n",
       "      <td>-3.276640</td>\n",
       "      <td>4.234669</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>2.115958</td>\n",
       "      <td>33.727214</td>\n",
       "      <td>-6.885687</td>\n",
       "      <td>-17.331922</td>\n",
       "      <td>-3.782329</td>\n",
       "      <td>-0.225454</td>\n",
       "      <td>-4.071704</td>\n",
       "      <td>19.412535</td>\n",
       "      <td>-36.365485</td>\n",
       "      <td>33.435179</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.574805</td>\n",
       "      <td>0.482555</td>\n",
       "      <td>-10.429128</td>\n",
       "      <td>2.159477</td>\n",
       "      <td>2.476564</td>\n",
       "      <td>-5.285379</td>\n",
       "      <td>2.654657</td>\n",
       "      <td>0.013844</td>\n",
       "      <td>1.870174</td>\n",
       "      <td>-5.366900</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>44.615872</td>\n",
       "      <td>238.765545</td>\n",
       "      <td>-34.731908</td>\n",
       "      <td>-20.485934</td>\n",
       "      <td>-54.539781</td>\n",
       "      <td>-13.233798</td>\n",
       "      <td>43.268734</td>\n",
       "      <td>-69.221238</td>\n",
       "      <td>-54.098555</td>\n",
       "      <td>103.271829</td>\n",
       "      <td>...</td>\n",
       "      <td>-10.187373</td>\n",
       "      <td>-28.663651</td>\n",
       "      <td>-111.895904</td>\n",
       "      <td>139.740226</td>\n",
       "      <td>119.098853</td>\n",
       "      <td>11.962286</td>\n",
       "      <td>-118.077865</td>\n",
       "      <td>-98.370769</td>\n",
       "      <td>-233.502787</td>\n",
       "      <td>-4.404923</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>24.974215</td>\n",
       "      <td>306.611650</td>\n",
       "      <td>-43.491294</td>\n",
       "      <td>-61.429688</td>\n",
       "      <td>-72.406019</td>\n",
       "      <td>-8.729750</td>\n",
       "      <td>9.672090</td>\n",
       "      <td>-42.564949</td>\n",
       "      <td>-18.624017</td>\n",
       "      <td>142.228319</td>\n",
       "      <td>...</td>\n",
       "      <td>1.314000</td>\n",
       "      <td>-6.075282</td>\n",
       "      <td>-36.137239</td>\n",
       "      <td>17.323573</td>\n",
       "      <td>56.021319</td>\n",
       "      <td>29.783794</td>\n",
       "      <td>31.427323</td>\n",
       "      <td>30.806316</td>\n",
       "      <td>-47.404275</td>\n",
       "      <td>-7.449075</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>17.446335</td>\n",
       "      <td>185.411116</td>\n",
       "      <td>-46.281376</td>\n",
       "      <td>-17.285736</td>\n",
       "      <td>-52.538440</td>\n",
       "      <td>-4.640911</td>\n",
       "      <td>-3.999988</td>\n",
       "      <td>-64.535718</td>\n",
       "      <td>78.684272</td>\n",
       "      <td>-63.500771</td>\n",
       "      <td>...</td>\n",
       "      <td>-12.296796</td>\n",
       "      <td>-11.121154</td>\n",
       "      <td>-46.178636</td>\n",
       "      <td>-20.019563</td>\n",
       "      <td>-37.176795</td>\n",
       "      <td>-34.775948</td>\n",
       "      <td>-45.139998</td>\n",
       "      <td>24.459670</td>\n",
       "      <td>-14.903830</td>\n",
       "      <td>36.199110</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>215.372032</td>\n",
       "      <td>2789.983184</td>\n",
       "      <td>-636.415677</td>\n",
       "      <td>-339.720653</td>\n",
       "      <td>-715.504788</td>\n",
       "      <td>-234.140411</td>\n",
       "      <td>2.625892</td>\n",
       "      <td>-2360.662353</td>\n",
       "      <td>-4486.007601</td>\n",
       "      <td>-6318.694458</td>\n",
       "      <td>...</td>\n",
       "      <td>-93.653419</td>\n",
       "      <td>-37.254008</td>\n",
       "      <td>-813.704363</td>\n",
       "      <td>1543.089919</td>\n",
       "      <td>21.459500</td>\n",
       "      <td>-9.670961</td>\n",
       "      <td>277.584839</td>\n",
       "      <td>-474.344080</td>\n",
       "      <td>-86.460589</td>\n",
       "      <td>-394.607608</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>37.216346</td>\n",
       "      <td>370.461743</td>\n",
       "      <td>-154.095418</td>\n",
       "      <td>440.477153</td>\n",
       "      <td>-69.755479</td>\n",
       "      <td>-23.771068</td>\n",
       "      <td>29.506285</td>\n",
       "      <td>-523.887724</td>\n",
       "      <td>395.295403</td>\n",
       "      <td>36.443657</td>\n",
       "      <td>...</td>\n",
       "      <td>-6.788656</td>\n",
       "      <td>9.558440</td>\n",
       "      <td>37.027489</td>\n",
       "      <td>452.491068</td>\n",
       "      <td>131.358239</td>\n",
       "      <td>364.439783</td>\n",
       "      <td>174.581431</td>\n",
       "      <td>-77.965364</td>\n",
       "      <td>-151.552011</td>\n",
       "      <td>213.311706</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20</th>\n",
       "      <td>0.281514</td>\n",
       "      <td>2.328073</td>\n",
       "      <td>-0.450640</td>\n",
       "      <td>0.023937</td>\n",
       "      <td>-0.523639</td>\n",
       "      <td>-0.147825</td>\n",
       "      <td>0.282349</td>\n",
       "      <td>-1.050785</td>\n",
       "      <td>-1.015248</td>\n",
       "      <td>0.968105</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.167097</td>\n",
       "      <td>-0.125950</td>\n",
       "      <td>-0.702004</td>\n",
       "      <td>1.039410</td>\n",
       "      <td>1.340899</td>\n",
       "      <td>-0.022496</td>\n",
       "      <td>-1.641817</td>\n",
       "      <td>-0.498988</td>\n",
       "      <td>-1.772378</td>\n",
       "      <td>-0.179083</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>6.371277</td>\n",
       "      <td>44.092308</td>\n",
       "      <td>-6.807660</td>\n",
       "      <td>-18.303567</td>\n",
       "      <td>-5.964485</td>\n",
       "      <td>4.586019</td>\n",
       "      <td>-2.900713</td>\n",
       "      <td>28.611200</td>\n",
       "      <td>36.124052</td>\n",
       "      <td>-22.093184</td>\n",
       "      <td>...</td>\n",
       "      <td>6.760986</td>\n",
       "      <td>14.244667</td>\n",
       "      <td>-3.988319</td>\n",
       "      <td>-1.093048</td>\n",
       "      <td>13.597649</td>\n",
       "      <td>5.585938</td>\n",
       "      <td>3.845947</td>\n",
       "      <td>0.439355</td>\n",
       "      <td>-12.998186</td>\n",
       "      <td>7.427078</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22</th>\n",
       "      <td>255.950520</td>\n",
       "      <td>43.580484</td>\n",
       "      <td>-9.311427</td>\n",
       "      <td>18.795891</td>\n",
       "      <td>-21.473272</td>\n",
       "      <td>-6.304337</td>\n",
       "      <td>15.152318</td>\n",
       "      <td>-76.359514</td>\n",
       "      <td>-6.952221</td>\n",
       "      <td>-3.980002</td>\n",
       "      <td>...</td>\n",
       "      <td>-12.342324</td>\n",
       "      <td>13.538845</td>\n",
       "      <td>38.051425</td>\n",
       "      <td>1.252863</td>\n",
       "      <td>-24.666171</td>\n",
       "      <td>8.336308</td>\n",
       "      <td>9.163659</td>\n",
       "      <td>1.509811</td>\n",
       "      <td>20.960269</td>\n",
       "      <td>11.385728</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23</th>\n",
       "      <td>117.783225</td>\n",
       "      <td>1287.714253</td>\n",
       "      <td>-214.968444</td>\n",
       "      <td>-412.776080</td>\n",
       "      <td>-210.310595</td>\n",
       "      <td>-49.776553</td>\n",
       "      <td>-0.669138</td>\n",
       "      <td>98.080131</td>\n",
       "      <td>-1206.238653</td>\n",
       "      <td>731.800162</td>\n",
       "      <td>...</td>\n",
       "      <td>-37.352020</td>\n",
       "      <td>-19.662641</td>\n",
       "      <td>-310.441094</td>\n",
       "      <td>414.898876</td>\n",
       "      <td>259.732975</td>\n",
       "      <td>-29.528260</td>\n",
       "      <td>-105.440926</td>\n",
       "      <td>-130.804089</td>\n",
       "      <td>-247.571890</td>\n",
       "      <td>-156.271756</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24</th>\n",
       "      <td>2.888623</td>\n",
       "      <td>25.201275</td>\n",
       "      <td>-4.245535</td>\n",
       "      <td>-3.477839</td>\n",
       "      <td>-5.818679</td>\n",
       "      <td>-1.267583</td>\n",
       "      <td>1.797164</td>\n",
       "      <td>-7.160785</td>\n",
       "      <td>-11.029967</td>\n",
       "      <td>12.698137</td>\n",
       "      <td>...</td>\n",
       "      <td>0.788185</td>\n",
       "      <td>0.222424</td>\n",
       "      <td>2.970832</td>\n",
       "      <td>3.382556</td>\n",
       "      <td>10.657756</td>\n",
       "      <td>0.302026</td>\n",
       "      <td>-0.857731</td>\n",
       "      <td>1.067713</td>\n",
       "      <td>-3.828600</td>\n",
       "      <td>1.265472</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25</th>\n",
       "      <td>85.200220</td>\n",
       "      <td>1194.610933</td>\n",
       "      <td>-223.905010</td>\n",
       "      <td>-613.914045</td>\n",
       "      <td>371.534794</td>\n",
       "      <td>-5.459041</td>\n",
       "      <td>-95.520397</td>\n",
       "      <td>445.328068</td>\n",
       "      <td>-841.959118</td>\n",
       "      <td>726.593315</td>\n",
       "      <td>...</td>\n",
       "      <td>109.881308</td>\n",
       "      <td>105.750913</td>\n",
       "      <td>249.965418</td>\n",
       "      <td>-45.340977</td>\n",
       "      <td>-249.987612</td>\n",
       "      <td>251.939472</td>\n",
       "      <td>127.140623</td>\n",
       "      <td>-435.657429</td>\n",
       "      <td>468.524512</td>\n",
       "      <td>-385.031966</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26</th>\n",
       "      <td>1.169457</td>\n",
       "      <td>17.249733</td>\n",
       "      <td>-3.121078</td>\n",
       "      <td>-9.177856</td>\n",
       "      <td>-2.775998</td>\n",
       "      <td>1.645292</td>\n",
       "      <td>-2.367166</td>\n",
       "      <td>17.385072</td>\n",
       "      <td>20.999588</td>\n",
       "      <td>-9.019122</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.571184</td>\n",
       "      <td>-0.182538</td>\n",
       "      <td>0.298422</td>\n",
       "      <td>-0.535126</td>\n",
       "      <td>1.872936</td>\n",
       "      <td>-0.444871</td>\n",
       "      <td>-0.821766</td>\n",
       "      <td>2.954579</td>\n",
       "      <td>-2.354258</td>\n",
       "      <td>1.567920</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>27</th>\n",
       "      <td>0.015321</td>\n",
       "      <td>0.222941</td>\n",
       "      <td>-0.010974</td>\n",
       "      <td>-0.113234</td>\n",
       "      <td>-0.016640</td>\n",
       "      <td>0.018406</td>\n",
       "      <td>-0.037664</td>\n",
       "      <td>0.193460</td>\n",
       "      <td>0.188970</td>\n",
       "      <td>-0.063901</td>\n",
       "      <td>...</td>\n",
       "      <td>0.082828</td>\n",
       "      <td>0.101918</td>\n",
       "      <td>0.173922</td>\n",
       "      <td>0.105071</td>\n",
       "      <td>-0.168076</td>\n",
       "      <td>0.040175</td>\n",
       "      <td>0.093096</td>\n",
       "      <td>-0.532971</td>\n",
       "      <td>0.477366</td>\n",
       "      <td>-0.367245</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>28</th>\n",
       "      <td>0.105055</td>\n",
       "      <td>1.540775</td>\n",
       "      <td>-0.225154</td>\n",
       "      <td>-0.727699</td>\n",
       "      <td>-0.177795</td>\n",
       "      <td>0.309902</td>\n",
       "      <td>-0.104019</td>\n",
       "      <td>0.968435</td>\n",
       "      <td>0.656557</td>\n",
       "      <td>-0.290924</td>\n",
       "      <td>...</td>\n",
       "      <td>1.639032</td>\n",
       "      <td>2.469325</td>\n",
       "      <td>0.868875</td>\n",
       "      <td>-0.143292</td>\n",
       "      <td>-0.148827</td>\n",
       "      <td>-0.222249</td>\n",
       "      <td>0.382193</td>\n",
       "      <td>-1.104702</td>\n",
       "      <td>1.184421</td>\n",
       "      <td>-0.449049</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>29</th>\n",
       "      <td>79.778438</td>\n",
       "      <td>979.528267</td>\n",
       "      <td>-177.210153</td>\n",
       "      <td>-448.217135</td>\n",
       "      <td>-135.168704</td>\n",
       "      <td>68.812687</td>\n",
       "      <td>-65.184425</td>\n",
       "      <td>593.875602</td>\n",
       "      <td>62.043230</td>\n",
       "      <td>220.875222</td>\n",
       "      <td>...</td>\n",
       "      <td>173.082249</td>\n",
       "      <td>329.556579</td>\n",
       "      <td>161.102457</td>\n",
       "      <td>277.120309</td>\n",
       "      <td>-207.898291</td>\n",
       "      <td>45.793834</td>\n",
       "      <td>-217.202917</td>\n",
       "      <td>-94.459984</td>\n",
       "      <td>-163.903785</td>\n",
       "      <td>173.671481</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>356711</th>\n",
       "      <td>3.714645</td>\n",
       "      <td>38.298513</td>\n",
       "      <td>-7.350215</td>\n",
       "      <td>-10.565996</td>\n",
       "      <td>-8.136129</td>\n",
       "      <td>-0.729079</td>\n",
       "      <td>-1.365988</td>\n",
       "      <td>-1.448198</td>\n",
       "      <td>-3.950267</td>\n",
       "      <td>-2.901430</td>\n",
       "      <td>...</td>\n",
       "      <td>-2.645150</td>\n",
       "      <td>-2.105847</td>\n",
       "      <td>-4.892094</td>\n",
       "      <td>3.391173</td>\n",
       "      <td>-49.350698</td>\n",
       "      <td>39.161918</td>\n",
       "      <td>-27.545974</td>\n",
       "      <td>23.567198</td>\n",
       "      <td>-0.948074</td>\n",
       "      <td>1.010792</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>356712</th>\n",
       "      <td>0.123348</td>\n",
       "      <td>1.655261</td>\n",
       "      <td>-0.147134</td>\n",
       "      <td>-0.684323</td>\n",
       "      <td>-0.213825</td>\n",
       "      <td>0.968023</td>\n",
       "      <td>0.015600</td>\n",
       "      <td>0.848167</td>\n",
       "      <td>0.910059</td>\n",
       "      <td>-0.201649</td>\n",
       "      <td>...</td>\n",
       "      <td>4.270399</td>\n",
       "      <td>8.017744</td>\n",
       "      <td>-0.064481</td>\n",
       "      <td>-0.102706</td>\n",
       "      <td>-0.261109</td>\n",
       "      <td>-0.239528</td>\n",
       "      <td>0.155735</td>\n",
       "      <td>-1.200327</td>\n",
       "      <td>1.183235</td>\n",
       "      <td>-0.565780</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>356713</th>\n",
       "      <td>35.091637</td>\n",
       "      <td>432.439983</td>\n",
       "      <td>-83.704053</td>\n",
       "      <td>-113.626257</td>\n",
       "      <td>-103.832358</td>\n",
       "      <td>13.521383</td>\n",
       "      <td>-23.268158</td>\n",
       "      <td>77.225503</td>\n",
       "      <td>428.161244</td>\n",
       "      <td>-161.486087</td>\n",
       "      <td>...</td>\n",
       "      <td>7.372315</td>\n",
       "      <td>6.480732</td>\n",
       "      <td>-112.049708</td>\n",
       "      <td>94.489778</td>\n",
       "      <td>-25.756632</td>\n",
       "      <td>15.214241</td>\n",
       "      <td>118.952355</td>\n",
       "      <td>50.231370</td>\n",
       "      <td>57.166858</td>\n",
       "      <td>-27.360209</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>356714</th>\n",
       "      <td>0.076762</td>\n",
       "      <td>0.698528</td>\n",
       "      <td>-0.034076</td>\n",
       "      <td>-0.189110</td>\n",
       "      <td>-0.031376</td>\n",
       "      <td>-0.018850</td>\n",
       "      <td>-0.007442</td>\n",
       "      <td>0.072682</td>\n",
       "      <td>-0.474186</td>\n",
       "      <td>0.281822</td>\n",
       "      <td>...</td>\n",
       "      <td>0.113395</td>\n",
       "      <td>0.053637</td>\n",
       "      <td>0.583476</td>\n",
       "      <td>-0.425161</td>\n",
       "      <td>0.367329</td>\n",
       "      <td>-0.050271</td>\n",
       "      <td>0.282889</td>\n",
       "      <td>-0.485148</td>\n",
       "      <td>0.215984</td>\n",
       "      <td>0.118598</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>356715</th>\n",
       "      <td>0.354712</td>\n",
       "      <td>5.040435</td>\n",
       "      <td>-0.537191</td>\n",
       "      <td>-2.324902</td>\n",
       "      <td>0.189773</td>\n",
       "      <td>0.181007</td>\n",
       "      <td>-0.305483</td>\n",
       "      <td>2.853892</td>\n",
       "      <td>1.456568</td>\n",
       "      <td>1.251539</td>\n",
       "      <td>...</td>\n",
       "      <td>-1.092015</td>\n",
       "      <td>-0.719352</td>\n",
       "      <td>-11.465674</td>\n",
       "      <td>-2.750741</td>\n",
       "      <td>1.146772</td>\n",
       "      <td>-2.506357</td>\n",
       "      <td>0.103743</td>\n",
       "      <td>5.508372</td>\n",
       "      <td>-0.909026</td>\n",
       "      <td>-0.582893</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>356716</th>\n",
       "      <td>6.169303</td>\n",
       "      <td>86.321139</td>\n",
       "      <td>-8.632180</td>\n",
       "      <td>-39.975357</td>\n",
       "      <td>-15.117626</td>\n",
       "      <td>6.052271</td>\n",
       "      <td>-11.182546</td>\n",
       "      <td>71.689844</td>\n",
       "      <td>84.646403</td>\n",
       "      <td>-28.739204</td>\n",
       "      <td>...</td>\n",
       "      <td>-8.278659</td>\n",
       "      <td>-8.837841</td>\n",
       "      <td>-5.446769</td>\n",
       "      <td>-7.073782</td>\n",
       "      <td>8.525366</td>\n",
       "      <td>-2.913883</td>\n",
       "      <td>-5.797928</td>\n",
       "      <td>19.780171</td>\n",
       "      <td>-13.896467</td>\n",
       "      <td>13.867128</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>356717</th>\n",
       "      <td>15.896641</td>\n",
       "      <td>225.031046</td>\n",
       "      <td>-33.363501</td>\n",
       "      <td>-99.751872</td>\n",
       "      <td>-16.916023</td>\n",
       "      <td>-2.845249</td>\n",
       "      <td>-18.663868</td>\n",
       "      <td>88.048751</td>\n",
       "      <td>-191.239401</td>\n",
       "      <td>158.246928</td>\n",
       "      <td>...</td>\n",
       "      <td>23.264607</td>\n",
       "      <td>10.838487</td>\n",
       "      <td>128.842462</td>\n",
       "      <td>-16.887814</td>\n",
       "      <td>23.677050</td>\n",
       "      <td>2.841492</td>\n",
       "      <td>32.635568</td>\n",
       "      <td>-134.313963</td>\n",
       "      <td>52.490962</td>\n",
       "      <td>25.088719</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>356718</th>\n",
       "      <td>2.574833</td>\n",
       "      <td>16.430907</td>\n",
       "      <td>-5.793350</td>\n",
       "      <td>10.420768</td>\n",
       "      <td>-3.654005</td>\n",
       "      <td>-1.182069</td>\n",
       "      <td>1.051799</td>\n",
       "      <td>-21.541827</td>\n",
       "      <td>9.557679</td>\n",
       "      <td>-3.948557</td>\n",
       "      <td>...</td>\n",
       "      <td>0.760273</td>\n",
       "      <td>-0.025464</td>\n",
       "      <td>9.449200</td>\n",
       "      <td>2.291437</td>\n",
       "      <td>11.750864</td>\n",
       "      <td>8.672937</td>\n",
       "      <td>2.176595</td>\n",
       "      <td>-3.045623</td>\n",
       "      <td>-6.905088</td>\n",
       "      <td>10.172591</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>356719</th>\n",
       "      <td>3.525007</td>\n",
       "      <td>34.644889</td>\n",
       "      <td>-11.415117</td>\n",
       "      <td>16.881748</td>\n",
       "      <td>-10.701201</td>\n",
       "      <td>-2.027451</td>\n",
       "      <td>2.364216</td>\n",
       "      <td>-43.165738</td>\n",
       "      <td>37.170977</td>\n",
       "      <td>-6.550927</td>\n",
       "      <td>...</td>\n",
       "      <td>1.113775</td>\n",
       "      <td>2.323951</td>\n",
       "      <td>6.198223</td>\n",
       "      <td>21.690214</td>\n",
       "      <td>6.389227</td>\n",
       "      <td>18.410111</td>\n",
       "      <td>8.331588</td>\n",
       "      <td>-10.748599</td>\n",
       "      <td>-0.690958</td>\n",
       "      <td>3.700754</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>356720</th>\n",
       "      <td>14.418895</td>\n",
       "      <td>191.393626</td>\n",
       "      <td>-21.652004</td>\n",
       "      <td>-36.655760</td>\n",
       "      <td>-41.534689</td>\n",
       "      <td>-10.984574</td>\n",
       "      <td>-9.834455</td>\n",
       "      <td>-91.665943</td>\n",
       "      <td>-246.294479</td>\n",
       "      <td>-259.034171</td>\n",
       "      <td>...</td>\n",
       "      <td>15.609286</td>\n",
       "      <td>7.224774</td>\n",
       "      <td>81.568434</td>\n",
       "      <td>43.106843</td>\n",
       "      <td>19.367209</td>\n",
       "      <td>-39.984017</td>\n",
       "      <td>50.151556</td>\n",
       "      <td>-6.336557</td>\n",
       "      <td>42.172235</td>\n",
       "      <td>12.724220</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>356721</th>\n",
       "      <td>0.906581</td>\n",
       "      <td>10.094002</td>\n",
       "      <td>-3.639514</td>\n",
       "      <td>7.356338</td>\n",
       "      <td>-2.854358</td>\n",
       "      <td>-0.798697</td>\n",
       "      <td>0.961450</td>\n",
       "      <td>-14.974028</td>\n",
       "      <td>8.815876</td>\n",
       "      <td>-1.143465</td>\n",
       "      <td>...</td>\n",
       "      <td>0.147569</td>\n",
       "      <td>0.281773</td>\n",
       "      <td>5.736391</td>\n",
       "      <td>2.524290</td>\n",
       "      <td>3.308487</td>\n",
       "      <td>5.006598</td>\n",
       "      <td>-2.756076</td>\n",
       "      <td>-1.809987</td>\n",
       "      <td>-2.111008</td>\n",
       "      <td>2.393843</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>356722</th>\n",
       "      <td>117.410751</td>\n",
       "      <td>288.980488</td>\n",
       "      <td>-32.527424</td>\n",
       "      <td>-87.825081</td>\n",
       "      <td>-20.620969</td>\n",
       "      <td>-7.368544</td>\n",
       "      <td>21.373037</td>\n",
       "      <td>31.695066</td>\n",
       "      <td>-88.786078</td>\n",
       "      <td>84.454278</td>\n",
       "      <td>...</td>\n",
       "      <td>99.436101</td>\n",
       "      <td>39.441211</td>\n",
       "      <td>888.068627</td>\n",
       "      <td>450.970751</td>\n",
       "      <td>-117.885655</td>\n",
       "      <td>-157.490396</td>\n",
       "      <td>33.545003</td>\n",
       "      <td>108.891756</td>\n",
       "      <td>-219.992054</td>\n",
       "      <td>-1.632823</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>356723</th>\n",
       "      <td>1.962753</td>\n",
       "      <td>11.345984</td>\n",
       "      <td>-2.895482</td>\n",
       "      <td>-2.339258</td>\n",
       "      <td>-2.635629</td>\n",
       "      <td>0.067351</td>\n",
       "      <td>-0.589343</td>\n",
       "      <td>-0.690461</td>\n",
       "      <td>7.932406</td>\n",
       "      <td>-5.503001</td>\n",
       "      <td>...</td>\n",
       "      <td>0.677126</td>\n",
       "      <td>0.722078</td>\n",
       "      <td>-1.048185</td>\n",
       "      <td>2.199398</td>\n",
       "      <td>-4.436782</td>\n",
       "      <td>0.640233</td>\n",
       "      <td>2.590578</td>\n",
       "      <td>-6.357509</td>\n",
       "      <td>6.713262</td>\n",
       "      <td>-3.451656</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>356724</th>\n",
       "      <td>27.623999</td>\n",
       "      <td>345.294795</td>\n",
       "      <td>-52.649894</td>\n",
       "      <td>-151.782583</td>\n",
       "      <td>-16.615234</td>\n",
       "      <td>-1.471138</td>\n",
       "      <td>-12.841293</td>\n",
       "      <td>156.411107</td>\n",
       "      <td>-166.981836</td>\n",
       "      <td>179.077411</td>\n",
       "      <td>...</td>\n",
       "      <td>-97.060500</td>\n",
       "      <td>72.764629</td>\n",
       "      <td>232.389182</td>\n",
       "      <td>57.227838</td>\n",
       "      <td>-107.923657</td>\n",
       "      <td>-7.859845</td>\n",
       "      <td>10.013677</td>\n",
       "      <td>-324.516025</td>\n",
       "      <td>158.493982</td>\n",
       "      <td>20.145142</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>356725</th>\n",
       "      <td>32.422199</td>\n",
       "      <td>328.753735</td>\n",
       "      <td>-96.563073</td>\n",
       "      <td>109.957451</td>\n",
       "      <td>-113.365177</td>\n",
       "      <td>-19.715226</td>\n",
       "      <td>21.616440</td>\n",
       "      <td>-401.030296</td>\n",
       "      <td>373.442941</td>\n",
       "      <td>-48.560163</td>\n",
       "      <td>...</td>\n",
       "      <td>0.204746</td>\n",
       "      <td>7.459718</td>\n",
       "      <td>28.382643</td>\n",
       "      <td>355.797376</td>\n",
       "      <td>98.313478</td>\n",
       "      <td>274.924399</td>\n",
       "      <td>87.896091</td>\n",
       "      <td>-86.446236</td>\n",
       "      <td>-122.113437</td>\n",
       "      <td>56.634143</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>356726</th>\n",
       "      <td>0.004867</td>\n",
       "      <td>0.053101</td>\n",
       "      <td>-0.019183</td>\n",
       "      <td>0.051086</td>\n",
       "      <td>-0.019011</td>\n",
       "      <td>-0.006289</td>\n",
       "      <td>0.011883</td>\n",
       "      <td>-0.118777</td>\n",
       "      <td>0.085768</td>\n",
       "      <td>0.045322</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.000824</td>\n",
       "      <td>0.002658</td>\n",
       "      <td>0.003884</td>\n",
       "      <td>0.087038</td>\n",
       "      <td>0.043369</td>\n",
       "      <td>0.070162</td>\n",
       "      <td>0.061657</td>\n",
       "      <td>-0.012519</td>\n",
       "      <td>-0.028697</td>\n",
       "      <td>0.042680</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>356727</th>\n",
       "      <td>46.702599</td>\n",
       "      <td>107.962040</td>\n",
       "      <td>-24.876417</td>\n",
       "      <td>15.938991</td>\n",
       "      <td>-31.557791</td>\n",
       "      <td>-7.091357</td>\n",
       "      <td>15.723770</td>\n",
       "      <td>-83.369972</td>\n",
       "      <td>24.880263</td>\n",
       "      <td>4.195042</td>\n",
       "      <td>...</td>\n",
       "      <td>-4.261053</td>\n",
       "      <td>-16.506046</td>\n",
       "      <td>-113.960619</td>\n",
       "      <td>-37.438808</td>\n",
       "      <td>-3.917616</td>\n",
       "      <td>-35.401697</td>\n",
       "      <td>75.905426</td>\n",
       "      <td>-0.645610</td>\n",
       "      <td>6.531441</td>\n",
       "      <td>-77.595594</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>356728</th>\n",
       "      <td>4.560592</td>\n",
       "      <td>69.102694</td>\n",
       "      <td>-7.884326</td>\n",
       "      <td>-30.695591</td>\n",
       "      <td>-4.015676</td>\n",
       "      <td>0.166075</td>\n",
       "      <td>-7.373009</td>\n",
       "      <td>27.458001</td>\n",
       "      <td>-45.809266</td>\n",
       "      <td>32.662359</td>\n",
       "      <td>...</td>\n",
       "      <td>12.893403</td>\n",
       "      <td>12.027322</td>\n",
       "      <td>60.515378</td>\n",
       "      <td>-13.777672</td>\n",
       "      <td>31.569091</td>\n",
       "      <td>3.594519</td>\n",
       "      <td>24.793753</td>\n",
       "      <td>-37.895583</td>\n",
       "      <td>20.657860</td>\n",
       "      <td>-2.477337</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>356729</th>\n",
       "      <td>0.595424</td>\n",
       "      <td>3.494652</td>\n",
       "      <td>-0.649225</td>\n",
       "      <td>-0.072384</td>\n",
       "      <td>-0.849612</td>\n",
       "      <td>-0.119034</td>\n",
       "      <td>0.217910</td>\n",
       "      <td>-1.607625</td>\n",
       "      <td>0.855898</td>\n",
       "      <td>-0.585324</td>\n",
       "      <td>...</td>\n",
       "      <td>0.732926</td>\n",
       "      <td>0.039024</td>\n",
       "      <td>-0.117324</td>\n",
       "      <td>-2.127056</td>\n",
       "      <td>0.035215</td>\n",
       "      <td>-1.244641</td>\n",
       "      <td>1.622156</td>\n",
       "      <td>-0.518799</td>\n",
       "      <td>1.393931</td>\n",
       "      <td>1.305640</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>356730</th>\n",
       "      <td>31.357031</td>\n",
       "      <td>189.668965</td>\n",
       "      <td>-7.122392</td>\n",
       "      <td>3.854348</td>\n",
       "      <td>-52.392238</td>\n",
       "      <td>-17.133748</td>\n",
       "      <td>88.778131</td>\n",
       "      <td>-85.846571</td>\n",
       "      <td>-16.848180</td>\n",
       "      <td>81.327736</td>\n",
       "      <td>...</td>\n",
       "      <td>-45.419300</td>\n",
       "      <td>-58.162897</td>\n",
       "      <td>-269.033747</td>\n",
       "      <td>279.848554</td>\n",
       "      <td>401.491550</td>\n",
       "      <td>62.447408</td>\n",
       "      <td>-293.089614</td>\n",
       "      <td>-21.668057</td>\n",
       "      <td>-464.435438</td>\n",
       "      <td>-52.899063</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>356731</th>\n",
       "      <td>35.959182</td>\n",
       "      <td>375.189973</td>\n",
       "      <td>-172.201423</td>\n",
       "      <td>580.376297</td>\n",
       "      <td>-19.748919</td>\n",
       "      <td>-16.606773</td>\n",
       "      <td>17.786713</td>\n",
       "      <td>-410.612702</td>\n",
       "      <td>308.016860</td>\n",
       "      <td>38.351250</td>\n",
       "      <td>...</td>\n",
       "      <td>-1.821610</td>\n",
       "      <td>-2.866740</td>\n",
       "      <td>65.990245</td>\n",
       "      <td>324.197570</td>\n",
       "      <td>23.335688</td>\n",
       "      <td>475.479338</td>\n",
       "      <td>77.232408</td>\n",
       "      <td>6.815901</td>\n",
       "      <td>-150.034566</td>\n",
       "      <td>152.132860</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>356732</th>\n",
       "      <td>9.422549</td>\n",
       "      <td>72.406071</td>\n",
       "      <td>-15.071092</td>\n",
       "      <td>-1.135598</td>\n",
       "      <td>-18.204619</td>\n",
       "      <td>-3.342457</td>\n",
       "      <td>5.999800</td>\n",
       "      <td>-40.288022</td>\n",
       "      <td>3.448625</td>\n",
       "      <td>9.024369</td>\n",
       "      <td>...</td>\n",
       "      <td>3.840171</td>\n",
       "      <td>8.608227</td>\n",
       "      <td>-0.245249</td>\n",
       "      <td>-28.024607</td>\n",
       "      <td>-9.482089</td>\n",
       "      <td>-28.520888</td>\n",
       "      <td>34.218381</td>\n",
       "      <td>-8.053195</td>\n",
       "      <td>35.974597</td>\n",
       "      <td>-16.293996</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>356733</th>\n",
       "      <td>30.324860</td>\n",
       "      <td>354.428941</td>\n",
       "      <td>-25.271255</td>\n",
       "      <td>-169.726469</td>\n",
       "      <td>8.151217</td>\n",
       "      <td>-0.164888</td>\n",
       "      <td>44.160131</td>\n",
       "      <td>236.150825</td>\n",
       "      <td>-59.737623</td>\n",
       "      <td>149.606733</td>\n",
       "      <td>...</td>\n",
       "      <td>11.214297</td>\n",
       "      <td>-1.800035</td>\n",
       "      <td>479.697803</td>\n",
       "      <td>359.560984</td>\n",
       "      <td>-110.975276</td>\n",
       "      <td>-141.003042</td>\n",
       "      <td>-117.995738</td>\n",
       "      <td>67.016616</td>\n",
       "      <td>-188.241283</td>\n",
       "      <td>49.870037</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>356734</th>\n",
       "      <td>3.697928</td>\n",
       "      <td>46.131876</td>\n",
       "      <td>-5.137457</td>\n",
       "      <td>-15.135801</td>\n",
       "      <td>-10.335766</td>\n",
       "      <td>0.868409</td>\n",
       "      <td>-0.365278</td>\n",
       "      <td>12.564354</td>\n",
       "      <td>10.091707</td>\n",
       "      <td>13.409920</td>\n",
       "      <td>...</td>\n",
       "      <td>10.344972</td>\n",
       "      <td>1.230180</td>\n",
       "      <td>-9.525433</td>\n",
       "      <td>-1.989615</td>\n",
       "      <td>-7.850131</td>\n",
       "      <td>-8.757364</td>\n",
       "      <td>0.886732</td>\n",
       "      <td>-4.765382</td>\n",
       "      <td>11.206172</td>\n",
       "      <td>-3.737928</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>356735</th>\n",
       "      <td>10.713605</td>\n",
       "      <td>147.722352</td>\n",
       "      <td>21.246388</td>\n",
       "      <td>-44.575118</td>\n",
       "      <td>-20.727958</td>\n",
       "      <td>-1.182372</td>\n",
       "      <td>-25.750643</td>\n",
       "      <td>27.240823</td>\n",
       "      <td>-109.721572</td>\n",
       "      <td>99.469962</td>\n",
       "      <td>...</td>\n",
       "      <td>5.832698</td>\n",
       "      <td>2.458837</td>\n",
       "      <td>-3.423236</td>\n",
       "      <td>-51.333327</td>\n",
       "      <td>26.757148</td>\n",
       "      <td>-28.947389</td>\n",
       "      <td>25.928053</td>\n",
       "      <td>-5.169472</td>\n",
       "      <td>22.601193</td>\n",
       "      <td>-6.312365</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>356736</th>\n",
       "      <td>9.925635</td>\n",
       "      <td>129.043114</td>\n",
       "      <td>-19.430378</td>\n",
       "      <td>-65.632607</td>\n",
       "      <td>9.274785</td>\n",
       "      <td>0.508033</td>\n",
       "      <td>-12.630326</td>\n",
       "      <td>69.302522</td>\n",
       "      <td>-80.599045</td>\n",
       "      <td>66.655813</td>\n",
       "      <td>...</td>\n",
       "      <td>15.406005</td>\n",
       "      <td>15.550583</td>\n",
       "      <td>170.375260</td>\n",
       "      <td>40.746979</td>\n",
       "      <td>-14.828934</td>\n",
       "      <td>47.409715</td>\n",
       "      <td>3.135798</td>\n",
       "      <td>-96.706523</td>\n",
       "      <td>30.350761</td>\n",
       "      <td>30.523790</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>356737</th>\n",
       "      <td>21.297139</td>\n",
       "      <td>136.550086</td>\n",
       "      <td>-31.535157</td>\n",
       "      <td>-5.130921</td>\n",
       "      <td>-23.748934</td>\n",
       "      <td>-3.666737</td>\n",
       "      <td>0.646709</td>\n",
       "      <td>-38.754607</td>\n",
       "      <td>6.474971</td>\n",
       "      <td>-26.655856</td>\n",
       "      <td>...</td>\n",
       "      <td>5.228506</td>\n",
       "      <td>3.447661</td>\n",
       "      <td>16.360908</td>\n",
       "      <td>-46.905805</td>\n",
       "      <td>-8.604090</td>\n",
       "      <td>-23.317906</td>\n",
       "      <td>-9.202495</td>\n",
       "      <td>-29.382524</td>\n",
       "      <td>11.029217</td>\n",
       "      <td>30.716404</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>356738</th>\n",
       "      <td>10.365171</td>\n",
       "      <td>79.529081</td>\n",
       "      <td>-15.348355</td>\n",
       "      <td>4.736660</td>\n",
       "      <td>-24.901692</td>\n",
       "      <td>-4.578115</td>\n",
       "      <td>4.807503</td>\n",
       "      <td>-59.730005</td>\n",
       "      <td>17.960198</td>\n",
       "      <td>3.006006</td>\n",
       "      <td>...</td>\n",
       "      <td>5.096534</td>\n",
       "      <td>-1.831885</td>\n",
       "      <td>-1.681517</td>\n",
       "      <td>-57.873213</td>\n",
       "      <td>-33.248660</td>\n",
       "      <td>-35.703996</td>\n",
       "      <td>9.794781</td>\n",
       "      <td>-13.095373</td>\n",
       "      <td>44.055192</td>\n",
       "      <td>-15.036217</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>356739</th>\n",
       "      <td>12.763440</td>\n",
       "      <td>139.517632</td>\n",
       "      <td>-43.220355</td>\n",
       "      <td>83.373885</td>\n",
       "      <td>-30.612257</td>\n",
       "      <td>-5.055345</td>\n",
       "      <td>4.624152</td>\n",
       "      <td>-121.301310</td>\n",
       "      <td>130.754234</td>\n",
       "      <td>1.049679</td>\n",
       "      <td>...</td>\n",
       "      <td>-2.874593</td>\n",
       "      <td>0.689334</td>\n",
       "      <td>-20.008722</td>\n",
       "      <td>79.897766</td>\n",
       "      <td>41.188507</td>\n",
       "      <td>51.751190</td>\n",
       "      <td>77.436445</td>\n",
       "      <td>25.140225</td>\n",
       "      <td>-23.769253</td>\n",
       "      <td>57.602697</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>356740</th>\n",
       "      <td>2.345557</td>\n",
       "      <td>27.483397</td>\n",
       "      <td>-4.603127</td>\n",
       "      <td>-11.967877</td>\n",
       "      <td>7.104591</td>\n",
       "      <td>-0.583996</td>\n",
       "      <td>-0.039055</td>\n",
       "      <td>5.439440</td>\n",
       "      <td>-15.020801</td>\n",
       "      <td>14.614613</td>\n",
       "      <td>...</td>\n",
       "      <td>2.531213</td>\n",
       "      <td>0.307054</td>\n",
       "      <td>8.625679</td>\n",
       "      <td>-4.102048</td>\n",
       "      <td>-4.110037</td>\n",
       "      <td>-0.969398</td>\n",
       "      <td>7.927082</td>\n",
       "      <td>-24.271058</td>\n",
       "      <td>16.877749</td>\n",
       "      <td>-11.643133</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>356741 rows × 40 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                 0            1           2           3           4  \\\n",
       "0        25.484256   190.556608  -31.338465  -15.281125  -45.671815   \n",
       "1       136.531465  1683.418578 -245.665873 -706.399956 -113.498056   \n",
       "2         0.847519    13.238973    0.447347   -3.803376    0.409061   \n",
       "3         0.774074     8.552813   -2.787724    5.090695   -2.393326   \n",
       "4         7.409898   103.713331  -15.019988  -47.338064   -8.188895   \n",
       "5       105.730569  1557.161201 -305.962548 -736.052351 -218.394935   \n",
       "6        29.349910   274.144686  -83.792972  131.760430  -92.864841   \n",
       "7         3.198312    36.748252   25.376836    9.704492   -8.451919   \n",
       "8         0.067702     1.186290    1.295021    0.007708   -0.190129   \n",
       "9        71.061732   256.103567  -48.734459   10.349713  -81.659392   \n",
       "10       57.222351   673.416804   32.139426 -215.983043  -77.710093   \n",
       "11        9.393029   124.085309   17.094020  -48.434145  -15.518081   \n",
       "12        0.812545     8.432800   -0.739086   -1.982202   -1.787722   \n",
       "13        2.821769    40.690979   49.265773    6.736184   -7.020680   \n",
       "14        2.115958    33.727214   -6.885687  -17.331922   -3.782329   \n",
       "15       44.615872   238.765545  -34.731908  -20.485934  -54.539781   \n",
       "16       24.974215   306.611650  -43.491294  -61.429688  -72.406019   \n",
       "17       17.446335   185.411116  -46.281376  -17.285736  -52.538440   \n",
       "18      215.372032  2789.983184 -636.415677 -339.720653 -715.504788   \n",
       "19       37.216346   370.461743 -154.095418  440.477153  -69.755479   \n",
       "20        0.281514     2.328073   -0.450640    0.023937   -0.523639   \n",
       "21        6.371277    44.092308   -6.807660  -18.303567   -5.964485   \n",
       "22      255.950520    43.580484   -9.311427   18.795891  -21.473272   \n",
       "23      117.783225  1287.714253 -214.968444 -412.776080 -210.310595   \n",
       "24        2.888623    25.201275   -4.245535   -3.477839   -5.818679   \n",
       "25       85.200220  1194.610933 -223.905010 -613.914045  371.534794   \n",
       "26        1.169457    17.249733   -3.121078   -9.177856   -2.775998   \n",
       "27        0.015321     0.222941   -0.010974   -0.113234   -0.016640   \n",
       "28        0.105055     1.540775   -0.225154   -0.727699   -0.177795   \n",
       "29       79.778438   979.528267 -177.210153 -448.217135 -135.168704   \n",
       "...            ...          ...         ...         ...         ...   \n",
       "356711    3.714645    38.298513   -7.350215  -10.565996   -8.136129   \n",
       "356712    0.123348     1.655261   -0.147134   -0.684323   -0.213825   \n",
       "356713   35.091637   432.439983  -83.704053 -113.626257 -103.832358   \n",
       "356714    0.076762     0.698528   -0.034076   -0.189110   -0.031376   \n",
       "356715    0.354712     5.040435   -0.537191   -2.324902    0.189773   \n",
       "356716    6.169303    86.321139   -8.632180  -39.975357  -15.117626   \n",
       "356717   15.896641   225.031046  -33.363501  -99.751872  -16.916023   \n",
       "356718    2.574833    16.430907   -5.793350   10.420768   -3.654005   \n",
       "356719    3.525007    34.644889  -11.415117   16.881748  -10.701201   \n",
       "356720   14.418895   191.393626  -21.652004  -36.655760  -41.534689   \n",
       "356721    0.906581    10.094002   -3.639514    7.356338   -2.854358   \n",
       "356722  117.410751   288.980488  -32.527424  -87.825081  -20.620969   \n",
       "356723    1.962753    11.345984   -2.895482   -2.339258   -2.635629   \n",
       "356724   27.623999   345.294795  -52.649894 -151.782583  -16.615234   \n",
       "356725   32.422199   328.753735  -96.563073  109.957451 -113.365177   \n",
       "356726    0.004867     0.053101   -0.019183    0.051086   -0.019011   \n",
       "356727   46.702599   107.962040  -24.876417   15.938991  -31.557791   \n",
       "356728    4.560592    69.102694   -7.884326  -30.695591   -4.015676   \n",
       "356729    0.595424     3.494652   -0.649225   -0.072384   -0.849612   \n",
       "356730   31.357031   189.668965   -7.122392    3.854348  -52.392238   \n",
       "356731   35.959182   375.189973 -172.201423  580.376297  -19.748919   \n",
       "356732    9.422549    72.406071  -15.071092   -1.135598  -18.204619   \n",
       "356733   30.324860   354.428941  -25.271255 -169.726469    8.151217   \n",
       "356734    3.697928    46.131876   -5.137457  -15.135801  -10.335766   \n",
       "356735   10.713605   147.722352   21.246388  -44.575118  -20.727958   \n",
       "356736    9.925635   129.043114  -19.430378  -65.632607    9.274785   \n",
       "356737   21.297139   136.550086  -31.535157   -5.130921  -23.748934   \n",
       "356738   10.365171    79.529081  -15.348355    4.736660  -24.901692   \n",
       "356739   12.763440   139.517632  -43.220355   83.373885  -30.612257   \n",
       "356740    2.345557    27.483397   -4.603127  -11.967877    7.104591   \n",
       "\n",
       "                 5           6            7            8            9  \\\n",
       "0        -5.856145    6.885757   -67.207197    33.708774    19.864555   \n",
       "1       -15.900123  -17.033764   633.092502  -698.810449  1206.443617   \n",
       "2        -0.157565   -1.905522     0.135612    -8.502080    -4.277642   \n",
       "3        -0.638501    0.677728   -10.803769     5.212503    -0.490907   \n",
       "4        -0.580236  -12.552942    45.456363   -89.877750    57.267064   \n",
       "5         6.163846 -167.804834   852.814889  -960.758184   908.356908   \n",
       "6       -20.080335   29.475516  -379.396561   299.740749    70.651411   \n",
       "7        -1.218361   -2.712137   -31.168713    15.768606    -1.285726   \n",
       "8        -0.174863    1.653260     0.010459     0.038666     0.359853   \n",
       "9       -20.028944   53.286094  -194.765781    -1.964136    47.967224   \n",
       "10       -4.450906  -58.813590   164.617646  -287.319744   411.922621   \n",
       "11       -2.077937   70.038885    90.884239    67.141310   -24.036977   \n",
       "12       -0.252183    1.456411    -0.118746    -2.127982     1.341633   \n",
       "13       -5.993160  127.343407     4.896962     3.774986     9.026917   \n",
       "14       -0.225454   -4.071704    19.412535   -36.365485    33.435179   \n",
       "15      -13.233798   43.268734   -69.221238   -54.098555   103.271829   \n",
       "16       -8.729750    9.672090   -42.564949   -18.624017   142.228319   \n",
       "17       -4.640911   -3.999988   -64.535718    78.684272   -63.500771   \n",
       "18     -234.140411    2.625892 -2360.662353 -4486.007601 -6318.694458   \n",
       "19      -23.771068   29.506285  -523.887724   395.295403    36.443657   \n",
       "20       -0.147825    0.282349    -1.050785    -1.015248     0.968105   \n",
       "21        4.586019   -2.900713    28.611200    36.124052   -22.093184   \n",
       "22       -6.304337   15.152318   -76.359514    -6.952221    -3.980002   \n",
       "23      -49.776553   -0.669138    98.080131 -1206.238653   731.800162   \n",
       "24       -1.267583    1.797164    -7.160785   -11.029967    12.698137   \n",
       "25       -5.459041  -95.520397   445.328068  -841.959118   726.593315   \n",
       "26        1.645292   -2.367166    17.385072    20.999588    -9.019122   \n",
       "27        0.018406   -0.037664     0.193460     0.188970    -0.063901   \n",
       "28        0.309902   -0.104019     0.968435     0.656557    -0.290924   \n",
       "29       68.812687  -65.184425   593.875602    62.043230   220.875222   \n",
       "...            ...         ...          ...          ...          ...   \n",
       "356711   -0.729079   -1.365988    -1.448198    -3.950267    -2.901430   \n",
       "356712    0.968023    0.015600     0.848167     0.910059    -0.201649   \n",
       "356713   13.521383  -23.268158    77.225503   428.161244  -161.486087   \n",
       "356714   -0.018850   -0.007442     0.072682    -0.474186     0.281822   \n",
       "356715    0.181007   -0.305483     2.853892     1.456568     1.251539   \n",
       "356716    6.052271  -11.182546    71.689844    84.646403   -28.739204   \n",
       "356717   -2.845249  -18.663868    88.048751  -191.239401   158.246928   \n",
       "356718   -1.182069    1.051799   -21.541827     9.557679    -3.948557   \n",
       "356719   -2.027451    2.364216   -43.165738    37.170977    -6.550927   \n",
       "356720  -10.984574   -9.834455   -91.665943  -246.294479  -259.034171   \n",
       "356721   -0.798697    0.961450   -14.974028     8.815876    -1.143465   \n",
       "356722   -7.368544   21.373037    31.695066   -88.786078    84.454278   \n",
       "356723    0.067351   -0.589343    -0.690461     7.932406    -5.503001   \n",
       "356724   -1.471138  -12.841293   156.411107  -166.981836   179.077411   \n",
       "356725  -19.715226   21.616440  -401.030296   373.442941   -48.560163   \n",
       "356726   -0.006289    0.011883    -0.118777     0.085768     0.045322   \n",
       "356727   -7.091357   15.723770   -83.369972    24.880263     4.195042   \n",
       "356728    0.166075   -7.373009    27.458001   -45.809266    32.662359   \n",
       "356729   -0.119034    0.217910    -1.607625     0.855898    -0.585324   \n",
       "356730  -17.133748   88.778131   -85.846571   -16.848180    81.327736   \n",
       "356731  -16.606773   17.786713  -410.612702   308.016860    38.351250   \n",
       "356732   -3.342457    5.999800   -40.288022     3.448625     9.024369   \n",
       "356733   -0.164888   44.160131   236.150825   -59.737623   149.606733   \n",
       "356734    0.868409   -0.365278    12.564354    10.091707    13.409920   \n",
       "356735   -1.182372  -25.750643    27.240823  -109.721572    99.469962   \n",
       "356736    0.508033  -12.630326    69.302522   -80.599045    66.655813   \n",
       "356737   -3.666737    0.646709   -38.754607     6.474971   -26.655856   \n",
       "356738   -4.578115    4.807503   -59.730005    17.960198     3.006006   \n",
       "356739   -5.055345    4.624152  -121.301310   130.754234     1.049679   \n",
       "356740   -0.583996   -0.039055     5.439440   -15.020801    14.614613   \n",
       "\n",
       "           ...              30          31           32           33  \\\n",
       "0          ...       18.908646    2.617127   -25.639946  -140.907299   \n",
       "1          ...      119.943753   27.056856  1150.207759   402.881649   \n",
       "2          ...        3.126801    2.241860    21.314967    -8.181559   \n",
       "3          ...       -0.239463   -0.025838     3.569100     0.399720   \n",
       "4          ...       18.248195   11.618561    80.990458   -15.497598   \n",
       "5          ...      -26.392069  -23.784510   -70.650686    -5.380971   \n",
       "6          ...       -0.383982    2.354587    -2.417050   172.204439   \n",
       "7          ...        0.380364    2.389711    11.452508     3.563717   \n",
       "8          ...       -0.121605   -0.178907    -0.324101     0.524534   \n",
       "9          ...      -30.107370  -63.556273  -431.644919    51.788518   \n",
       "10         ...       34.912641  -16.006519   214.332887   -20.128511   \n",
       "11         ...       14.876476    4.032670   225.302064    69.192388   \n",
       "12         ...       33.565733  -16.045833     0.424453    -0.633023   \n",
       "13         ...       31.507425   56.899463    -8.361463     4.232159   \n",
       "14         ...       -0.574805    0.482555   -10.429128     2.159477   \n",
       "15         ...      -10.187373  -28.663651  -111.895904   139.740226   \n",
       "16         ...        1.314000   -6.075282   -36.137239    17.323573   \n",
       "17         ...      -12.296796  -11.121154   -46.178636   -20.019563   \n",
       "18         ...      -93.653419  -37.254008  -813.704363  1543.089919   \n",
       "19         ...       -6.788656    9.558440    37.027489   452.491068   \n",
       "20         ...       -0.167097   -0.125950    -0.702004     1.039410   \n",
       "21         ...        6.760986   14.244667    -3.988319    -1.093048   \n",
       "22         ...      -12.342324   13.538845    38.051425     1.252863   \n",
       "23         ...      -37.352020  -19.662641  -310.441094   414.898876   \n",
       "24         ...        0.788185    0.222424     2.970832     3.382556   \n",
       "25         ...      109.881308  105.750913   249.965418   -45.340977   \n",
       "26         ...       -0.571184   -0.182538     0.298422    -0.535126   \n",
       "27         ...        0.082828    0.101918     0.173922     0.105071   \n",
       "28         ...        1.639032    2.469325     0.868875    -0.143292   \n",
       "29         ...      173.082249  329.556579   161.102457   277.120309   \n",
       "...        ...             ...         ...          ...          ...   \n",
       "356711     ...       -2.645150   -2.105847    -4.892094     3.391173   \n",
       "356712     ...        4.270399    8.017744    -0.064481    -0.102706   \n",
       "356713     ...        7.372315    6.480732  -112.049708    94.489778   \n",
       "356714     ...        0.113395    0.053637     0.583476    -0.425161   \n",
       "356715     ...       -1.092015   -0.719352   -11.465674    -2.750741   \n",
       "356716     ...       -8.278659   -8.837841    -5.446769    -7.073782   \n",
       "356717     ...       23.264607   10.838487   128.842462   -16.887814   \n",
       "356718     ...        0.760273   -0.025464     9.449200     2.291437   \n",
       "356719     ...        1.113775    2.323951     6.198223    21.690214   \n",
       "356720     ...       15.609286    7.224774    81.568434    43.106843   \n",
       "356721     ...        0.147569    0.281773     5.736391     2.524290   \n",
       "356722     ...       99.436101   39.441211   888.068627   450.970751   \n",
       "356723     ...        0.677126    0.722078    -1.048185     2.199398   \n",
       "356724     ...      -97.060500   72.764629   232.389182    57.227838   \n",
       "356725     ...        0.204746    7.459718    28.382643   355.797376   \n",
       "356726     ...       -0.000824    0.002658     0.003884     0.087038   \n",
       "356727     ...       -4.261053  -16.506046  -113.960619   -37.438808   \n",
       "356728     ...       12.893403   12.027322    60.515378   -13.777672   \n",
       "356729     ...        0.732926    0.039024    -0.117324    -2.127056   \n",
       "356730     ...      -45.419300  -58.162897  -269.033747   279.848554   \n",
       "356731     ...       -1.821610   -2.866740    65.990245   324.197570   \n",
       "356732     ...        3.840171    8.608227    -0.245249   -28.024607   \n",
       "356733     ...       11.214297   -1.800035   479.697803   359.560984   \n",
       "356734     ...       10.344972    1.230180    -9.525433    -1.989615   \n",
       "356735     ...        5.832698    2.458837    -3.423236   -51.333327   \n",
       "356736     ...       15.406005   15.550583   170.375260    40.746979   \n",
       "356737     ...        5.228506    3.447661    16.360908   -46.905805   \n",
       "356738     ...        5.096534   -1.831885    -1.681517   -57.873213   \n",
       "356739     ...       -2.874593    0.689334   -20.008722    79.897766   \n",
       "356740     ...        2.531213    0.307054     8.625679    -4.102048   \n",
       "\n",
       "                34          35          36           37          38  \\\n",
       "0      -138.664991 -101.156227   83.021828   -60.092995  174.518318   \n",
       "1      -240.484972  282.666337  -11.156897 -1598.418982  413.945688   \n",
       "2        16.993917    2.947127    9.914219    -4.323199    2.386330   \n",
       "3         2.237289    1.984111   -3.765371     1.062166   -4.699627   \n",
       "4        21.382244   -5.109917   15.604298   -59.106464   29.079635   \n",
       "5       147.839625 -216.426469  -82.874996     1.231659 -137.057685   \n",
       "6         2.921642  116.451911   -8.540848   -45.587263  -54.867679   \n",
       "7         0.498055    7.156311   -6.503403    -5.383621   -0.662745   \n",
       "8        -0.726809   -0.414440   -0.419759    -0.019525   -0.000276   \n",
       "9       128.044258   38.745012  197.894843   111.665653 -266.072087   \n",
       "10      131.563741  106.336729   97.776656  -232.446729   48.091937   \n",
       "11       31.292021  -51.494772   19.965973   114.762513  -70.685876   \n",
       "12       -0.423473   -0.708624   -0.833846    -1.892525    1.415557   \n",
       "13        3.130651    5.321420    2.812752    -0.126850   -3.276640   \n",
       "14        2.476564   -5.285379    2.654657     0.013844    1.870174   \n",
       "15      119.098853   11.962286 -118.077865   -98.370769 -233.502787   \n",
       "16       56.021319   29.783794   31.427323    30.806316  -47.404275   \n",
       "17      -37.176795  -34.775948  -45.139998    24.459670  -14.903830   \n",
       "18       21.459500   -9.670961  277.584839  -474.344080  -86.460589   \n",
       "19      131.358239  364.439783  174.581431   -77.965364 -151.552011   \n",
       "20        1.340899   -0.022496   -1.641817    -0.498988   -1.772378   \n",
       "21       13.597649    5.585938    3.845947     0.439355  -12.998186   \n",
       "22      -24.666171    8.336308    9.163659     1.509811   20.960269   \n",
       "23      259.732975  -29.528260 -105.440926  -130.804089 -247.571890   \n",
       "24       10.657756    0.302026   -0.857731     1.067713   -3.828600   \n",
       "25     -249.987612  251.939472  127.140623  -435.657429  468.524512   \n",
       "26        1.872936   -0.444871   -0.821766     2.954579   -2.354258   \n",
       "27       -0.168076    0.040175    0.093096    -0.532971    0.477366   \n",
       "28       -0.148827   -0.222249    0.382193    -1.104702    1.184421   \n",
       "29     -207.898291   45.793834 -217.202917   -94.459984 -163.903785   \n",
       "...            ...         ...         ...          ...         ...   \n",
       "356711  -49.350698   39.161918  -27.545974    23.567198   -0.948074   \n",
       "356712   -0.261109   -0.239528    0.155735    -1.200327    1.183235   \n",
       "356713  -25.756632   15.214241  118.952355    50.231370   57.166858   \n",
       "356714    0.367329   -0.050271    0.282889    -0.485148    0.215984   \n",
       "356715    1.146772   -2.506357    0.103743     5.508372   -0.909026   \n",
       "356716    8.525366   -2.913883   -5.797928    19.780171  -13.896467   \n",
       "356717   23.677050    2.841492   32.635568  -134.313963   52.490962   \n",
       "356718   11.750864    8.672937    2.176595    -3.045623   -6.905088   \n",
       "356719    6.389227   18.410111    8.331588   -10.748599   -0.690958   \n",
       "356720   19.367209  -39.984017   50.151556    -6.336557   42.172235   \n",
       "356721    3.308487    5.006598   -2.756076    -1.809987   -2.111008   \n",
       "356722 -117.885655 -157.490396   33.545003   108.891756 -219.992054   \n",
       "356723   -4.436782    0.640233    2.590578    -6.357509    6.713262   \n",
       "356724 -107.923657   -7.859845   10.013677  -324.516025  158.493982   \n",
       "356725   98.313478  274.924399   87.896091   -86.446236 -122.113437   \n",
       "356726    0.043369    0.070162    0.061657    -0.012519   -0.028697   \n",
       "356727   -3.917616  -35.401697   75.905426    -0.645610    6.531441   \n",
       "356728   31.569091    3.594519   24.793753   -37.895583   20.657860   \n",
       "356729    0.035215   -1.244641    1.622156    -0.518799    1.393931   \n",
       "356730  401.491550   62.447408 -293.089614   -21.668057 -464.435438   \n",
       "356731   23.335688  475.479338   77.232408     6.815901 -150.034566   \n",
       "356732   -9.482089  -28.520888   34.218381    -8.053195   35.974597   \n",
       "356733 -110.975276 -141.003042 -117.995738    67.016616 -188.241283   \n",
       "356734   -7.850131   -8.757364    0.886732    -4.765382   11.206172   \n",
       "356735   26.757148  -28.947389   25.928053    -5.169472   22.601193   \n",
       "356736  -14.828934   47.409715    3.135798   -96.706523   30.350761   \n",
       "356737   -8.604090  -23.317906   -9.202495   -29.382524   11.029217   \n",
       "356738  -33.248660  -35.703996    9.794781   -13.095373   44.055192   \n",
       "356739   41.188507   51.751190   77.436445    25.140225  -23.769253   \n",
       "356740   -4.110037   -0.969398    7.927082   -24.271058   16.877749   \n",
       "\n",
       "                39  \n",
       "0        16.115740  \n",
       "1       -71.406816  \n",
       "2         2.616657  \n",
       "3         9.414097  \n",
       "4        22.248695  \n",
       "5       140.238067  \n",
       "6        26.263262  \n",
       "7       -13.626061  \n",
       "8         0.198532  \n",
       "9       -94.960629  \n",
       "10     -182.587256  \n",
       "11       -0.977672  \n",
       "12        2.027099  \n",
       "13        4.234669  \n",
       "14       -5.366900  \n",
       "15       -4.404923  \n",
       "16       -7.449075  \n",
       "17       36.199110  \n",
       "18     -394.607608  \n",
       "19      213.311706  \n",
       "20       -0.179083  \n",
       "21        7.427078  \n",
       "22       11.385728  \n",
       "23     -156.271756  \n",
       "24        1.265472  \n",
       "25     -385.031966  \n",
       "26        1.567920  \n",
       "27       -0.367245  \n",
       "28       -0.449049  \n",
       "29      173.671481  \n",
       "...            ...  \n",
       "356711    1.010792  \n",
       "356712   -0.565780  \n",
       "356713  -27.360209  \n",
       "356714    0.118598  \n",
       "356715   -0.582893  \n",
       "356716   13.867128  \n",
       "356717   25.088719  \n",
       "356718   10.172591  \n",
       "356719    3.700754  \n",
       "356720   12.724220  \n",
       "356721    2.393843  \n",
       "356722   -1.632823  \n",
       "356723   -3.451656  \n",
       "356724   20.145142  \n",
       "356725   56.634143  \n",
       "356726    0.042680  \n",
       "356727  -77.595594  \n",
       "356728   -2.477337  \n",
       "356729    1.305640  \n",
       "356730  -52.899063  \n",
       "356731  152.132860  \n",
       "356732  -16.293996  \n",
       "356733   49.870037  \n",
       "356734   -3.737928  \n",
       "356735   -6.312365  \n",
       "356736   30.523790  \n",
       "356737   30.716404  \n",
       "356738  -15.036217  \n",
       "356739   57.602697  \n",
       "356740  -11.643133  \n",
       "\n",
       "[356741 rows x 40 columns]"
      ]
     },
     "execution_count": 73,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cuml_result.to_pandas()"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.5.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}