מטלת MNIST.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "import seaborn as sns\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "df = pd.read_csv('MNIST/fashion-mnist_train.csv', dtype=int) # read train data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>label</th>\n",
       "      <th>pixel1</th>\n",
       "      <th>pixel2</th>\n",
       "      <th>pixel3</th>\n",
       "      <th>pixel4</th>\n",
       "      <th>pixel5</th>\n",
       "      <th>pixel6</th>\n",
       "      <th>pixel7</th>\n",
       "      <th>pixel8</th>\n",
       "      <th>pixel9</th>\n",
       "      <th>...</th>\n",
       "      <th>pixel775</th>\n",
       "      <th>pixel776</th>\n",
       "      <th>pixel777</th>\n",
       "      <th>pixel778</th>\n",
       "      <th>pixel779</th>\n",
       "      <th>pixel780</th>\n",
       "      <th>pixel781</th>\n",
       "      <th>pixel782</th>\n",
       "      <th>pixel783</th>\n",
       "      <th>pixel784</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</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>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>9</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</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>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>6</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>5</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>30</td>\n",
       "      <td>43</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0</td>\n",
       "      <td>0</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>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>0</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",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</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>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 785 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   label  pixel1  pixel2  pixel3  pixel4  pixel5  pixel6  pixel7  pixel8  \\\n",
       "0      2       0       0       0       0       0       0       0       0   \n",
       "1      9       0       0       0       0       0       0       0       0   \n",
       "2      6       0       0       0       0       0       0       0       5   \n",
       "3      0       0       0       0       1       2       0       0       0   \n",
       "4      3       0       0       0       0       0       0       0       0   \n",
       "\n",
       "   pixel9  ...  pixel775  pixel776  pixel777  pixel778  pixel779  pixel780  \\\n",
       "0       0  ...         0         0         0         0         0         0   \n",
       "1       0  ...         0         0         0         0         0         0   \n",
       "2       0  ...         0         0         0        30        43         0   \n",
       "3       0  ...         3         0         0         0         0         1   \n",
       "4       0  ...         0         0         0         0         0         0   \n",
       "\n",
       "   pixel781  pixel782  pixel783  pixel784  \n",
       "0         0         0         0         0  \n",
       "1         0         0         0         0  \n",
       "2         0         0         0         0  \n",
       "3         0         0         0         0  \n",
       "4         0         0         0         0  \n",
       "\n",
       "[5 rows x 785 columns]"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 60000 entries, 0 to 59999\n",
      "Columns: 785 entries, label to pixel784\n",
      "dtypes: int32(785)\n",
      "memory usage: 179.7 MB\n"
     ]
    }
   ],
   "source": [
    "df.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "9    6000\n",
       "8    6000\n",
       "7    6000\n",
       "6    6000\n",
       "5    6000\n",
       "4    6000\n",
       "3    6000\n",
       "2    6000\n",
       "1    6000\n",
       "0    6000\n",
       "Name: label, dtype: int64"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df['label'].value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.metrics import confusion_matrix,classification_report , accuracy_score"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[1 1 1 ... 1 1 1]\n"
     ]
    }
   ],
   "source": [
    "h_0 = np.full(60000,1)\n",
    "print(h_0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.00      0.00      0.00      6000\n",
      "           1       0.10      1.00      0.18      6000\n",
      "           2       0.00      0.00      0.00      6000\n",
      "           3       0.00      0.00      0.00      6000\n",
      "           4       0.00      0.00      0.00      6000\n",
      "           5       0.00      0.00      0.00      6000\n",
      "           6       0.00      0.00      0.00      6000\n",
      "           7       0.00      0.00      0.00      6000\n",
      "           8       0.00      0.00      0.00      6000\n",
      "           9       0.00      0.00      0.00      6000\n",
      "\n",
      "    accuracy                           0.10     60000\n",
      "   macro avg       0.01      0.10      0.02     60000\n",
      "weighted avg       0.01      0.10      0.02     60000\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\ProgramData\\Anaconda3\\lib\\site-packages\\sklearn\\metrics\\_classification.py:1272: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n"
     ]
    }
   ],
   "source": [
    "print(classification_report(df.label,h_0))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "בדיקת מודלים"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "X = df.drop(\"label\",axis = 1)\n",
    "y = df.label"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.ensemble import RandomForestClassifier"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "RandomForestClassifier(bootstrap=True, ccp_alpha=0.0, class_weight=None,\n",
       "                       criterion='gini', max_depth=None, max_features='auto',\n",
       "                       max_leaf_nodes=16, max_samples=None,\n",
       "                       min_impurity_decrease=0.0, min_impurity_split=None,\n",
       "                       min_samples_leaf=1, min_samples_split=2,\n",
       "                       min_weight_fraction_leaf=0.0, n_estimators=500,\n",
       "                       n_jobs=-1, oob_score=False, random_state=None, verbose=0,\n",
       "                       warm_start=False)"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rnd_clf = RandomForestClassifier(n_estimators=500, max_leaf_nodes=16, n_jobs=-1)\n",
    "rnd_clf.fit(X, y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "fi = rnd_clf.feature_importances_\n",
    "pi = np.reshape(fi,(28,28))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x576 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize= (10,8))\n",
    "ax = sns.heatmap(pi,cmap='BuPu',linewidths=.5)\n",
    "cbar = ax.collections[0].colorbar\n",
    "cbar.set_ticks([0,max(fi)])\n",
    "cbar.set_ticklabels(['not important','Most important'])\n",
    "plt.title('pixel importance',size = 20)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.decomposition import PCA\n",
    "pca= PCA(n_components=0.95)\n",
    "x_reduced = pca.fit_transform(X)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "GaussianNB(priors=None, var_smoothing=1e-09)\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.74      0.74      0.74      3056\n",
      "           1       0.94      0.88      0.91      2957\n",
      "           2       0.73      0.67      0.70      3009\n",
      "           3       0.79      0.76      0.78      2960\n",
      "           4       0.74      0.68      0.71      3060\n",
      "           5       0.80      0.73      0.76      3022\n",
      "           6       0.56      0.47      0.51      2960\n",
      "           7       0.74      0.85      0.79      3019\n",
      "           8       0.61      0.87      0.72      2985\n",
      "           9       0.93      0.86      0.89      2972\n",
      "\n",
      "    accuracy                           0.75     30000\n",
      "   macro avg       0.76      0.75      0.75     30000\n",
      "weighted avg       0.76      0.75      0.75     30000\n",
      "\n",
      "[[2268   14   49  170   13   32  178    0  332    0]\n",
      " [   9 2616   24  151    5   20   34    0   98    0]\n",
      " [  81    0 2009   16  345   30  247    1  280    0]\n",
      " [ 130  131   30 2260   85   33  119    0  172    0]\n",
      " [  17    9  302  121 2071    8  304    0  228    0]\n",
      " [  37    0   17    0    0 2195   90  564   94   25]\n",
      " [ 455    5  299  102  234   43 1403    0  419    0]\n",
      " [   2    0    0    0    0  280    4 2571    4  158]\n",
      " [  52    0   24   25   45   60   96   78 2601    4]\n",
      " [   6    0    1    1    0   48   32  266   60 2558]]\n"
     ]
    }
   ],
   "source": [
    "from sklearn import metrics\n",
    "from sklearn.naive_bayes import GaussianNB\n",
    "\n",
    "from sklearn.model_selection import train_test_split\n",
    "X_train, X_test, y_train, y_test = train_test_split(x_reduced, y, test_size=0.5, random_state=0)\n",
    "\n",
    "model = GaussianNB()\n",
    "\n",
    "model.fit(X_train, y_train).predict(X_test)\n",
    "print(model)\n",
    "expected = y_train\n",
    "predicted = model.predict(X_train)\n",
    "print(metrics.classification_report(expected, predicted))\n",
    "print(metrics.confusion_matrix(expected, predicted))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "def logRegModel(X, y):\n",
    "    # Split the data (X, Y)\n",
    "    from sklearn import metrics\n",
    "    X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=101)\n",
    "\n",
    "    # Create and train model\n",
    "    logmodel = LogisticRegression()\n",
    "    logmodel.fit(X_train, y_train)\n",
    "    predictions = logmodel.predict(X_test)\n",
    "\n",
    "    # Evaluate the model\n",
    "    cfm = metrics.confusion_matrix(y_test, predictions)\n",
    "    sns.heatmap(cfm, annot=True)\n",
    "    print(classification_report(y_test, predictions))\n",
    "    \n",
    "    \n",
    "def SVMModel(x, y):\n",
    "    # Split the data (X, Y)\n",
    "    from sklearn.model_selection import train_test_split\n",
    "    X_train, X_test, y_train, y_test = train_test_split(x, y, test_size = 0.20)\n",
    "    \n",
    "    # Create and train model\n",
    "    from sklearn.svm import SVC\n",
    "    svclassifier = SVC(kernel='linear')\n",
    "    svclassifier.fit(X_train, y_train)\n",
    "    y_pred = svclassifier.predict(X_test)\n",
    "    \n",
    "    # Evaluate the model\n",
    "    from sklearn.metrics import classification_report, confusion_matrix\n",
    "    cfm = confusion_matrix(y_test,y_pred)\n",
    "    sns.heatmap(cfm, annot=True)\n",
    "    print(classification_report(y_test,y_pred))\n",
    "    \n",
    "def knn(x, y):\n",
    "    # Split the data (X, Y)\n",
    "    from sklearn.model_selection import train_test_split\n",
    "    X_train, X_test, y_train, y_test = train_test_split(x, y, test_size=0.20)\n",
    "    \n",
    "    # Create and train model\n",
    "    from sklearn.preprocessing import StandardScaler\n",
    "    scaler = StandardScaler()\n",
    "    scaler.fit(X_train)\n",
    "    X_train = scaler.transform(X_train)\n",
    "    X_test = scaler.transform(X_test)\n",
    "    from sklearn.neighbors import KNeighborsClassifier\n",
    "    classifier = KNeighborsClassifier(n_neighbors=5)\n",
    "    classifier.fit(X_train, y_train)\n",
    "    y_pred = classifier.predict(X_test)\n",
    "    \n",
    "    # Evaluate the model\n",
    "    from sklearn.metrics import classification_report, confusion_matrix\n",
    "    cfm = confusion_matrix(y_test,y_pred)\n",
    "    sns.heatmap(cfm, annot=True)\n",
    "    print(classification_report(y_test, y_pred))\n",
    "\n",
    "def tree(x,y):\n",
    "    # Split the data (X, Y)\n",
    "    from sklearn.model_selection import train_test_split\n",
    "    X_train, X_test, y_train, y_test = train_test_split(x, y, test_size=0.20)\n",
    "    \n",
    "    # Create and train model\n",
    "    from sklearn.tree import DecisionTreeClassifier\n",
    "    classifier = DecisionTreeClassifier()\n",
    "    classifier.fit(X_train, y_train)\n",
    "    y_pred = classifier.predict(X_test)\n",
    "    \n",
    "    # Evaluate the model\n",
    "    from sklearn.metrics import classification_report, confusion_matrix\n",
    "    print(confusion_matrix(y_test, y_pred))\n",
    "    print(classification_report(y_test, y_pred))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[ 867    6   18   93   28    2  198    0   18    2]\n",
      " [   8 1158    7   44   12    1   10    0    0    0]\n",
      " [  26   10  723   26  186    2  156    0   29    1]\n",
      " [  73   37   15  927   55    2   60    0   20    1]\n",
      " [  19    7  181   60  727    5  144    0   25    1]\n",
      " [   3    3    3    2    3 1014    4  106   25   56]\n",
      " [ 179   11  151   53  145    2  611    1   31    0]\n",
      " [   2    0    0    0    0   98    0 1059    6   88]\n",
      " [  19    3   21   20   17   26   33    8 1032    6]\n",
      " [   1    0    0    1    1   56    0   82    7 1021]]\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.72      0.70      0.71      1232\n",
      "           1       0.94      0.93      0.94      1240\n",
      "           2       0.65      0.62      0.63      1159\n",
      "           3       0.76      0.78      0.77      1190\n",
      "           4       0.62      0.62      0.62      1169\n",
      "           5       0.84      0.83      0.84      1219\n",
      "           6       0.50      0.52      0.51      1184\n",
      "           7       0.84      0.85      0.84      1253\n",
      "           8       0.87      0.87      0.87      1185\n",
      "           9       0.87      0.87      0.87      1169\n",
      "\n",
      "    accuracy                           0.76     12000\n",
      "   macro avg       0.76      0.76      0.76     12000\n",
      "weighted avg       0.76      0.76      0.76     12000\n",
      "\n"
     ]
    }
   ],
   "source": [
    "tree(x_reduced,y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "SVMModel(x_reduced, y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "logRegModel(x_reduced,y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "df['label'].plot.pie()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "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": "3.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}