Time Series DataBlock-Loop.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "from fastai.vision import * #import from vision to use the classification interpreter\n",
    "from pathlib import Path\n",
    "import pdb\n",
    "import seaborn as sns\n",
    "from sklearn.decomposition import PCA\n",
    "from timeseries import TimeSeriesItem, TimeSeriesList, UCRArchive"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "ucr = UCRArchive()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "class TSClassifier(torch.nn.Module):\n",
    "    def __init__(self,seqLen,numClasses):\n",
    "        super().__init__()\n",
    "        self.conv = WideResNet1d(3,3,6)\n",
    "        self.hidden= torch.nn.Linear(self.conv.nf,300)\n",
    "        self.out = torch.nn.Linear(300,numClasses)\n",
    "              \n",
    "    def forward(self,ts):\n",
    "        ts = self.conv(ts.unsqueeze(1))\n",
    "        return self.out(torch.relu(self.hidden(ts)))\n",
    "    \n",
    "class TSAE(torch.nn.Module):\n",
    "    def __init__(self,seqLen,latentDim=24):\n",
    "        super().__init__()\n",
    "        #self.conv = ConvBase(seqLen)\n",
    "        self.conv = WideResNet1d(3,3,6)\n",
    "        self.mean = torch.nn.Linear(self.conv.nf,latentDim)\n",
    "        self.logvar = torch.nn.Linear(self.conv.nf,latentDim)\n",
    "        \n",
    "\n",
    "        layers = []\n",
    "        for a,b in [(latentDim,100),(100,200),(200,300)]:\n",
    "            layers += bn_drop_lin(a,b,actn=torch.nn.ReLU())\n",
    "        self.lin = torch.nn.Sequential(*layers)\n",
    "        self.out = torch.nn.Linear(300,seqLen)\n",
    "\n",
    "    def forward(self,ts):\n",
    "        seqLen = ts.shape[1]\n",
    "        ts = self.conv(ts.unsqueeze(1)).squeeze(-1)\n",
    "        #mean, logvar = self.mean(ts), self.logvar(ts)\n",
    "        mean = self.mean(ts)\n",
    "        \n",
    "        ls = mean\n",
    "#         if self.training:\n",
    "#             std = torch.exp(0.5 * logvar)\n",
    "#             eps = torch.randn_like(std)\n",
    "#             ls = eps.mul(std).add_(mean)\n",
    "        return self.out(self.lin(ls))          \n",
    "        #return self.out(self.lin(ls)), mean, logvar"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "def EvaluateSeries(dataset_name):\n",
    "    out = [dataset_name]\n",
    "    print(f\"Evaluating {dataset_name}\")\n",
    "    try:\n",
    "        data = (TimeSeriesList.from_csv_list(ucr.get_csv_files(dataset_name),labelCol=0)\n",
    "                          .split_by_csv_name([f\"{dataset_name}_TEST.tsv\"])\n",
    "                          .label_from_col())\n",
    "        dslen = min(len(data.train),len(data.valid))\n",
    "        bs = min(64,dslen//10)\n",
    "        data = data.databunch(num_workers=0,bs=bs)\n",
    "\n",
    "        dataAE = (TimeSeriesList.from_csv_list(ucr.get_csv_files(dataset_name),labelCol=0)\n",
    "                            .no_split()\n",
    "                            .label_from_self()\n",
    "                            .databunch(num_workers=0,bs=bs))\n",
    "\n",
    "        learn = Learner(data,TSClassifier(len(data.train_ds[0][0].data),data.train_ds.c),loss_func=F.cross_entropy,metrics=[accuracy])\n",
    "        learn.fit_one_cycle(10,1e-3,wd=0.2)\n",
    "        learn.fit_one_cycle(10,1e-4,wd=0.2)\n",
    "        out.append(max([m[0].item() for m in learn.recorder.metrics]))\n",
    "\n",
    "        learnAE = Learner(dataAE,TSAE(len(data.train_ds[0][0].data)),loss_func=F.mse_loss)\n",
    "        learnAE.fit_one_cycle(10,1e-2)\n",
    "        learnAE.fit_one_cycle(10,1e-3)\n",
    "        out.append(learnAE.validate(dataAE.train_dl)[0])\n",
    "        \n",
    "        learnDT = Learner(data,TSClassifier(len(data.train_ds[0][0].data),data.train_ds.c),loss_func=F.cross_entropy,metrics=[accuracy],\n",
    "                 callback_fns=BnFreeze,bn_wd=False,train_bn=False)\n",
    "        learnDT.split(split_model(learnDT.model,[*learnDT.model.conv.split_groups,learnDT.model.out]))\n",
    "        learnDT.model.conv.load_state_dict(learnAE.model.conv.state_dict())\n",
    "        learnDT.freeze_to(-1)\n",
    "        learnDT.fit_one_cycle(10,1e-2)\n",
    "        out.append(max([m[0].item() for m in learnDT.recorder.metrics]))\n",
    "        learnDT.unfreeze()\n",
    "        learnDT.fit_one_cycle(10,slice(1e-5,1e-2))\n",
    "        out.append(max([m[0].item() for m in learnDT.recorder.metrics]))\n",
    "\n",
    "    except:\n",
    "        pass\n",
    "    \n",
    "    return out"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "Total time: 00:23 <p><table style='width:300px; margin-bottom:10px'>\n",
       "  <tr>\n",
       "    <th>epoch</th>\n",
       "    <th>train_loss</th>\n",
       "    <th>valid_loss</th>\n",
       "    <th>accuracy</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>1</th>\n",
       "    <th>1.610730</th>\n",
       "    <th>1.611045</th>\n",
       "    <th>0.129485</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>2</th>\n",
       "    <th>1.606811</th>\n",
       "    <th>1.584814</th>\n",
       "    <th>0.379381</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>3</th>\n",
       "    <th>1.592688</th>\n",
       "    <th>1.556604</th>\n",
       "    <th>0.255258</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>4</th>\n",
       "    <th>1.574267</th>\n",
       "    <th>1.495077</th>\n",
       "    <th>0.563299</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>5</th>\n",
       "    <th>1.553732</th>\n",
       "    <th>1.451006</th>\n",
       "    <th>0.571134</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>6</th>\n",
       "    <th>1.527374</th>\n",
       "    <th>1.433113</th>\n",
       "    <th>0.565361</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>7</th>\n",
       "    <th>1.503614</th>\n",
       "    <th>1.419905</th>\n",
       "    <th>0.566598</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>8</th>\n",
       "    <th>1.481314</th>\n",
       "    <th>1.402775</th>\n",
       "    <th>0.574433</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>9</th>\n",
       "    <th>1.462593</th>\n",
       "    <th>1.395341</th>\n",
       "    <th>0.587629</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>10</th>\n",
       "    <th>1.447530</th>\n",
       "    <th>1.394382</th>\n",
       "    <th>0.588454</th>\n",
       "  </tr>\n",
       "</table>\n"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "Total time: 00:26 <p><table style='width:300px; margin-bottom:10px'>\n",
       "  <tr>\n",
       "    <th>epoch</th>\n",
       "    <th>train_loss</th>\n",
       "    <th>valid_loss</th>\n",
       "    <th>accuracy</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>1</th>\n",
       "    <th>1.386247</th>\n",
       "    <th>1.305086</th>\n",
       "    <th>0.575670</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>2</th>\n",
       "    <th>1.368190</th>\n",
       "    <th>1.167847</th>\n",
       "    <th>0.594227</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>3</th>\n",
       "    <th>1.366220</th>\n",
       "    <th>1.413186</th>\n",
       "    <th>0.322474</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>4</th>\n",
       "    <th>1.263737</th>\n",
       "    <th>0.933581</th>\n",
       "    <th>0.640412</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>5</th>\n",
       "    <th>1.202673</th>\n",
       "    <th>0.904639</th>\n",
       "    <th>0.634639</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>6</th>\n",
       "    <th>1.131942</th>\n",
       "    <th>0.799438</th>\n",
       "    <th>0.672990</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>7</th>\n",
       "    <th>1.092678</th>\n",
       "    <th>0.780231</th>\n",
       "    <th>0.675052</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>8</th>\n",
       "    <th>1.045040</th>\n",
       "    <th>0.757951</th>\n",
       "    <th>0.763711</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>9</th>\n",
       "    <th>0.991999</th>\n",
       "    <th>0.756956</th>\n",
       "    <th>0.735670</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>10</th>\n",
       "    <th>0.939966</th>\n",
       "    <th>0.753049</th>\n",
       "    <th>0.727835</th>\n",
       "  </tr>\n",
       "</table>\n"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Evaluating MoteStrain\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "Total time: 00:56 <p><table style='width:300px; margin-bottom:10px'>\n",
       "  <tr>\n",
       "    <th>epoch</th>\n",
       "    <th>train_loss</th>\n",
       "    <th>valid_loss</th>\n",
       "    <th>accuracy</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>1</th>\n",
       "    <th>0.705079</th>\n",
       "    <th>0.697615</th>\n",
       "    <th>0.460863</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>2</th>\n",
       "    <th>0.703761</th>\n",
       "    <th>0.697559</th>\n",
       "    <th>0.460863</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>3</th>\n",
       "    <th>0.700654</th>\n",
       "    <th>0.689142</th>\n",
       "    <th>0.471246</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>4</th>\n",
       "    <th>0.664506</th>\n",
       "    <th>0.457358</th>\n",
       "    <th>0.828275</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>5</th>\n",
       "    <th>0.617854</th>\n",
       "    <th>0.687380</th>\n",
       "    <th>0.764377</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>6</th>\n",
       "    <th>0.647741</th>\n",
       "    <th>1.277048</th>\n",
       "    <th>0.598243</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>7</th>\n",
       "    <th>0.574782</th>\n",
       "    <th>0.745887</th>\n",
       "    <th>0.642173</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>8</th>\n",
       "    <th>0.576172</th>\n",
       "    <th>0.341902</th>\n",
       "    <th>0.853834</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>9</th>\n",
       "    <th>0.554631</th>\n",
       "    <th>0.326793</th>\n",
       "    <th>0.878594</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>10</th>\n",
       "    <th>0.493488</th>\n",
       "    <th>0.359921</th>\n",
       "    <th>0.850639</th>\n",
       "  </tr>\n",
       "</table>\n"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "Total time: 00:55 <p><table style='width:300px; margin-bottom:10px'>\n",
       "  <tr>\n",
       "    <th>epoch</th>\n",
       "    <th>train_loss</th>\n",
       "    <th>valid_loss</th>\n",
       "    <th>accuracy</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>1</th>\n",
       "    <th>0.352948</th>\n",
       "    <th>0.346560</th>\n",
       "    <th>0.861022</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>2</th>\n",
       "    <th>0.460151</th>\n",
       "    <th>0.332744</th>\n",
       "    <th>0.871406</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>3</th>\n",
       "    <th>0.407590</th>\n",
       "    <th>0.344868</th>\n",
       "    <th>0.854633</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>4</th>\n",
       "    <th>0.347824</th>\n",
       "    <th>0.360357</th>\n",
       "    <th>0.841054</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>5</th>\n",
       "    <th>0.339248</th>\n",
       "    <th>0.334879</th>\n",
       "    <th>0.859425</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>6</th>\n",
       "    <th>0.357742</th>\n",
       "    <th>0.326651</th>\n",
       "    <th>0.861821</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>7</th>\n",
       "    <th>0.345833</th>\n",
       "    <th>0.323189</th>\n",
       "    <th>0.865016</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>8</th>\n",
       "    <th>0.335817</th>\n",
       "    <th>0.328165</th>\n",
       "    <th>0.860224</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>9</th>\n",
       "    <th>0.355908</th>\n",
       "    <th>0.337448</th>\n",
       "    <th>0.854633</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>10</th>\n",
       "    <th>0.342101</th>\n",
       "    <th>0.331479</th>\n",
       "    <th>0.857827</th>\n",
       "  </tr>\n",
       "</table>\n"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "Total time: 04:59 <p><table style='width:300px; margin-bottom:10px'>\n",
       "  <tr>\n",
       "    <th>epoch</th>\n",
       "    <th>train_loss</th>\n",
       "    <th>valid_loss</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>1</th>\n",
       "    <th>0.564791</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>2</th>\n",
       "    <th>0.524218</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>3</th>\n",
       "    <th>0.509802</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>4</th>\n",
       "    <th>0.486679</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>5</th>\n",
       "    <th>0.528732</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>6</th>\n",
       "    <th>0.522147</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>7</th>\n",
       "    <th>0.507604</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>8</th>\n",
       "    <th>0.539938</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>9</th>\n",
       "    <th>0.511050</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>10</th>\n",
       "    <th>0.508767</th>\n",
       "  </tr>\n",
       "</table>\n"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "\n",
       "    <div>\n",
       "        <style>\n",
       "            /* Turns off some styling */\n",
       "            progress {\n",
       "                /* gets rid of default border in Firefox and Opera. */\n",
       "                border: none;\n",
       "                /* Needs to be in here for Safari polyfill so background images work as expected. */\n",
       "                background-size: auto;\n",
       "            }\n",
       "            .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n",
       "                background: #F44336;\n",
       "            }\n",
       "        </style>\n",
       "      <progress value='7' class='' max='10', style='width:300px; height:20px; vertical-align: middle;'></progress>\n",
       "      70.00% [7/10 03:30<01:30]\n",
       "    </div>\n",
       "    \n",
       "<table style='width:300px; margin-bottom:10px'>\n",
       "  <tr>\n",
       "    <th>epoch</th>\n",
       "    <th>train_loss</th>\n",
       "    <th>valid_loss</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>1</th>\n",
       "    <th>0.506482</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>2</th>\n",
       "    <th>0.523547</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>3</th>\n",
       "    <th>0.515189</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>4</th>\n",
       "    <th>0.505389</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>5</th>\n",
       "    <th>0.544122</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>6</th>\n",
       "    <th>0.512639</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>7</th>\n",
       "    <th>0.517005</th>\n",
       "  </tr>\n",
       "</table>\n",
       "\n",
       "\n",
       "    <div>\n",
       "        <style>\n",
       "            /* Turns off some styling */\n",
       "            progress {\n",
       "                /* gets rid of default border in Firefox and Opera. */\n",
       "                border: none;\n",
       "                /* Needs to be in here for Safari polyfill so background images work as expected. */\n",
       "                background-size: auto;\n",
       "            }\n",
       "            .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n",
       "                background: #F44336;\n",
       "            }\n",
       "        </style>\n",
       "      <progress value='0' class='progress-bar-interrupted' max='636', style='width:300px; height:20px; vertical-align: middle;'></progress>\n",
       "      Interrupted\n",
       "    </div>\n",
       "    "
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Evaluating NonInvasiveFetalECGThorax1\n",
      "Evaluating NonInvasiveFetalECGThorax2\n",
      "Evaluating OliveOil\n",
      "Evaluating OSULeaf\n",
      "Evaluating PhalangesOutlinesCorrect\n",
      "Evaluating Phoneme\n",
      "Evaluating PickupGestureWiimoteZ\n",
      "Evaluating PigAirwayPressure\n",
      "Evaluating PigArtPressure\n",
      "Evaluating PigCVP\n",
      "Evaluating PLAID\n",
      "Evaluating Plane\n",
      "Evaluating PowerCons\n",
      "Evaluating ProximalPhalanxOutlineAgeGroup\n",
      "Evaluating ProximalPhalanxOutlineCorrect\n",
      "Evaluating ProximalPhalanxTW\n",
      "Evaluating RefrigerationDevices\n",
      "Evaluating Rock\n",
      "Evaluating ScreenType\n",
      "Evaluating SemgHandGenderCh2\n",
      "Evaluating SemgHandMovementCh2\n",
      "Evaluating SemgHandSubjectCh2\n",
      "Evaluating ShakeGestureWiimoteZ\n",
      "Evaluating ShapeletSim\n",
      "Evaluating ShapesAll\n",
      "Evaluating SmallKitchenAppliances\n",
      "Evaluating SmoothSubspace\n",
      "Evaluating SonyAIBORobotSurface1\n",
      "Evaluating SonyAIBORobotSurface2\n",
      "Evaluating StarLightCurves\n",
      "Evaluating Strawberry\n",
      "Evaluating SwedishLeaf\n",
      "Evaluating Symbols\n",
      "Evaluating SyntheticControl\n",
      "Evaluating ToeSegmentation1\n",
      "Evaluating ToeSegmentation2\n",
      "Evaluating Trace\n",
      "Evaluating TwoLeadECG\n",
      "Evaluating TwoPatterns\n",
      "Evaluating UMD\n",
      "Evaluating UWaveGestureLibraryAll\n",
      "Evaluating UWaveGestureLibraryX\n",
      "Evaluating UWaveGestureLibraryY\n",
      "Evaluating UWaveGestureLibraryZ\n",
      "Evaluating Wafer\n",
      "Evaluating Wine\n",
      "Evaluating WordSynonyms\n",
      "Evaluating Worms\n",
      "Evaluating WormsTwoClass\n",
      "Evaluating Yoga\n"
     ]
    }
   ],
   "source": [
    "results = [EvaluateSeries(d) for d in ucr.list_datasets()]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "rDF = pd.DataFrame(results,columns=[\"Name\",\"Random\",\"AE Loss\",\"RandomCat\",\"Concat\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "rDF[\"GainT\"] = rDF[\"Concat\"] - rDF[\"Random\"]\n",
    "rDF[\"GainC\"] = rDF[\"Concat\"] - rDF[\"RandomCat\"]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "rDF.to_csv(\"ucrresults.csv\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "rDF = pd.read_csv(\"ucrresults.csv\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "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>Name</th>\n",
       "      <th>Random</th>\n",
       "      <th>AE Loss</th>\n",
       "      <th>RandomCat</th>\n",
       "      <th>Concat</th>\n",
       "      <th>GainT</th>\n",
       "      <th>GainC</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>69</th>\n",
       "      <td>Meat</td>\n",
       "      <td>0.333333</td>\n",
       "      <td>0.005748</td>\n",
       "      <td>0.950000</td>\n",
       "      <td>0.650000</td>\n",
       "      <td>0.316667</td>\n",
       "      <td>-0.300000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>108</th>\n",
       "      <td>Strawberry</td>\n",
       "      <td>0.586486</td>\n",
       "      <td>0.010888</td>\n",
       "      <td>0.959459</td>\n",
       "      <td>0.832432</td>\n",
       "      <td>0.245946</td>\n",
       "      <td>-0.127027</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>ArrowHead</td>\n",
       "      <td>0.548571</td>\n",
       "      <td>0.100859</td>\n",
       "      <td>0.748571</td>\n",
       "      <td>0.788571</td>\n",
       "      <td>0.240000</td>\n",
       "      <td>0.040000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>68</th>\n",
       "      <td>Mallat</td>\n",
       "      <td>0.776972</td>\n",
       "      <td>0.021429</td>\n",
       "      <td>0.939446</td>\n",
       "      <td>0.973134</td>\n",
       "      <td>0.196162</td>\n",
       "      <td>0.033689</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>73</th>\n",
       "      <td>MiddlePhalanxOutlineCorrect</td>\n",
       "      <td>0.570447</td>\n",
       "      <td>0.009673</td>\n",
       "      <td>0.800687</td>\n",
       "      <td>0.762887</td>\n",
       "      <td>0.192440</td>\n",
       "      <td>-0.037801</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>CricketZ</td>\n",
       "      <td>0.320513</td>\n",
       "      <td>0.484863</td>\n",
       "      <td>0.535897</td>\n",
       "      <td>0.502564</td>\n",
       "      <td>0.182051</td>\n",
       "      <td>-0.033333</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>92</th>\n",
       "      <td>ProximalPhalanxOutlineCorrect</td>\n",
       "      <td>0.646048</td>\n",
       "      <td>0.008137</td>\n",
       "      <td>0.890034</td>\n",
       "      <td>0.817869</td>\n",
       "      <td>0.171821</td>\n",
       "      <td>-0.072165</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>Coffee</td>\n",
       "      <td>0.857143</td>\n",
       "      <td>0.016528</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.142857</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>37</th>\n",
       "      <td>FaceFour</td>\n",
       "      <td>0.647727</td>\n",
       "      <td>0.734557</td>\n",
       "      <td>0.840909</td>\n",
       "      <td>0.784091</td>\n",
       "      <td>0.136364</td>\n",
       "      <td>-0.056818</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>Beef</td>\n",
       "      <td>0.466667</td>\n",
       "      <td>0.167091</td>\n",
       "      <td>0.800000</td>\n",
       "      <td>0.600000</td>\n",
       "      <td>0.133333</td>\n",
       "      <td>-0.200000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>127</th>\n",
       "      <td>Yoga</td>\n",
       "      <td>0.581333</td>\n",
       "      <td>0.027104</td>\n",
       "      <td>0.794000</td>\n",
       "      <td>0.710333</td>\n",
       "      <td>0.129000</td>\n",
       "      <td>-0.083667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>87</th>\n",
       "      <td>PigCVP</td>\n",
       "      <td>0.048077</td>\n",
       "      <td>4.197798</td>\n",
       "      <td>0.221154</td>\n",
       "      <td>0.173077</td>\n",
       "      <td>0.125000</td>\n",
       "      <td>-0.048077</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>51</th>\n",
       "      <td>GunPoint</td>\n",
       "      <td>0.813333</td>\n",
       "      <td>0.074187</td>\n",
       "      <td>0.960000</td>\n",
       "      <td>0.926667</td>\n",
       "      <td>0.113333</td>\n",
       "      <td>-0.033333</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>ACSF1</td>\n",
       "      <td>0.490000</td>\n",
       "      <td>0.144930</td>\n",
       "      <td>0.480000</td>\n",
       "      <td>0.600000</td>\n",
       "      <td>0.110000</td>\n",
       "      <td>0.120000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>115</th>\n",
       "      <td>TwoLeadECG</td>\n",
       "      <td>0.542581</td>\n",
       "      <td>0.141020</td>\n",
       "      <td>0.821773</td>\n",
       "      <td>0.651449</td>\n",
       "      <td>0.108867</td>\n",
       "      <td>-0.170325</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>CricketX</td>\n",
       "      <td>0.371795</td>\n",
       "      <td>0.486353</td>\n",
       "      <td>0.569231</td>\n",
       "      <td>0.479487</td>\n",
       "      <td>0.107692</td>\n",
       "      <td>-0.089744</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42</th>\n",
       "      <td>FordB</td>\n",
       "      <td>0.661728</td>\n",
       "      <td>0.770388</td>\n",
       "      <td>0.817284</td>\n",
       "      <td>0.769136</td>\n",
       "      <td>0.107407</td>\n",
       "      <td>-0.048148</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>86</th>\n",
       "      <td>PigArtPressure</td>\n",
       "      <td>0.072115</td>\n",
       "      <td>129.167374</td>\n",
       "      <td>0.201923</td>\n",
       "      <td>0.177885</td>\n",
       "      <td>0.105769</td>\n",
       "      <td>-0.024038</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>67</th>\n",
       "      <td>Lightning7</td>\n",
       "      <td>0.671233</td>\n",
       "      <td>0.484338</td>\n",
       "      <td>0.643836</td>\n",
       "      <td>0.767123</td>\n",
       "      <td>0.095890</td>\n",
       "      <td>0.123288</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Adiac</td>\n",
       "      <td>0.135550</td>\n",
       "      <td>0.013636</td>\n",
       "      <td>0.767263</td>\n",
       "      <td>0.227621</td>\n",
       "      <td>0.092072</td>\n",
       "      <td>-0.539642</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23</th>\n",
       "      <td>DistalPhalanxOutlineCorrect</td>\n",
       "      <td>0.612319</td>\n",
       "      <td>0.027200</td>\n",
       "      <td>0.782609</td>\n",
       "      <td>0.699275</td>\n",
       "      <td>0.086957</td>\n",
       "      <td>-0.083333</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>40</th>\n",
       "      <td>Fish</td>\n",
       "      <td>0.548571</td>\n",
       "      <td>0.028790</td>\n",
       "      <td>0.902857</td>\n",
       "      <td>0.634286</td>\n",
       "      <td>0.085714</td>\n",
       "      <td>-0.268571</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>93</th>\n",
       "      <td>ProximalPhalanxTW</td>\n",
       "      <td>0.731707</td>\n",
       "      <td>0.007223</td>\n",
       "      <td>0.804878</td>\n",
       "      <td>0.814634</td>\n",
       "      <td>0.082927</td>\n",
       "      <td>0.009756</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>96</th>\n",
       "      <td>ScreenType</td>\n",
       "      <td>0.357333</td>\n",
       "      <td>0.456331</td>\n",
       "      <td>0.354667</td>\n",
       "      <td>0.440000</td>\n",
       "      <td>0.082667</td>\n",
       "      <td>0.085333</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>114</th>\n",
       "      <td>Trace</td>\n",
       "      <td>0.580000</td>\n",
       "      <td>0.111137</td>\n",
       "      <td>0.940000</td>\n",
       "      <td>0.660000</td>\n",
       "      <td>0.080000</td>\n",
       "      <td>-0.280000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>81</th>\n",
       "      <td>OSULeaf</td>\n",
       "      <td>0.376033</td>\n",
       "      <td>0.471557</td>\n",
       "      <td>0.615702</td>\n",
       "      <td>0.446281</td>\n",
       "      <td>0.070248</td>\n",
       "      <td>-0.169421</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>82</th>\n",
       "      <td>PhalangesOutlinesCorrect</td>\n",
       "      <td>0.618881</td>\n",
       "      <td>0.015968</td>\n",
       "      <td>0.794872</td>\n",
       "      <td>0.688811</td>\n",
       "      <td>0.069930</td>\n",
       "      <td>-0.106061</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>112</th>\n",
       "      <td>ToeSegmentation1</td>\n",
       "      <td>0.552632</td>\n",
       "      <td>0.991212</td>\n",
       "      <td>0.657895</td>\n",
       "      <td>0.618421</td>\n",
       "      <td>0.065790</td>\n",
       "      <td>-0.039474</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>66</th>\n",
       "      <td>Lightning2</td>\n",
       "      <td>0.737705</td>\n",
       "      <td>0.490883</td>\n",
       "      <td>0.622951</td>\n",
       "      <td>0.803279</td>\n",
       "      <td>0.065574</td>\n",
       "      <td>0.180328</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>124</th>\n",
       "      <td>WordSynonyms</td>\n",
       "      <td>0.476489</td>\n",
       "      <td>0.199155</td>\n",
       "      <td>0.583072</td>\n",
       "      <td>0.540752</td>\n",
       "      <td>0.064263</td>\n",
       "      <td>-0.042320</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",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>116</th>\n",
       "      <td>TwoPatterns</td>\n",
       "      <td>0.960000</td>\n",
       "      <td>0.231215</td>\n",
       "      <td>0.999250</td>\n",
       "      <td>0.924500</td>\n",
       "      <td>-0.035500</td>\n",
       "      <td>-0.074750</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>33</th>\n",
       "      <td>EOGHorizontalSignal</td>\n",
       "      <td>0.477901</td>\n",
       "      <td>4723.121094</td>\n",
       "      <td>0.464088</td>\n",
       "      <td>0.439227</td>\n",
       "      <td>-0.038674</td>\n",
       "      <td>-0.024862</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>83</th>\n",
       "      <td>Phoneme</td>\n",
       "      <td>0.121308</td>\n",
       "      <td>0.883969</td>\n",
       "      <td>0.109705</td>\n",
       "      <td>0.078059</td>\n",
       "      <td>-0.043249</td>\n",
       "      <td>-0.031646</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>85</th>\n",
       "      <td>PigAirwayPressure</td>\n",
       "      <td>0.120192</td>\n",
       "      <td>2.420203</td>\n",
       "      <td>0.120192</td>\n",
       "      <td>0.072115</td>\n",
       "      <td>-0.048077</td>\n",
       "      <td>-0.048077</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>76</th>\n",
       "      <td>MixedShapesSmallTrain</td>\n",
       "      <td>0.842887</td>\n",
       "      <td>0.086877</td>\n",
       "      <td>0.838763</td>\n",
       "      <td>0.792990</td>\n",
       "      <td>-0.049897</td>\n",
       "      <td>-0.045773</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>32</th>\n",
       "      <td>ElectricDevices</td>\n",
       "      <td>0.714304</td>\n",
       "      <td>0.561187</td>\n",
       "      <td>0.694852</td>\n",
       "      <td>0.663468</td>\n",
       "      <td>-0.050837</td>\n",
       "      <td>-0.031384</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75</th>\n",
       "      <td>MixedShapesRegularTrain</td>\n",
       "      <td>0.896495</td>\n",
       "      <td>0.052715</td>\n",
       "      <td>0.922474</td>\n",
       "      <td>0.840000</td>\n",
       "      <td>-0.056495</td>\n",
       "      <td>-0.082474</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>59</th>\n",
       "      <td>HouseTwenty</td>\n",
       "      <td>0.722689</td>\n",
       "      <td>649137.250000</td>\n",
       "      <td>0.873950</td>\n",
       "      <td>0.663866</td>\n",
       "      <td>-0.058824</td>\n",
       "      <td>-0.210084</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>58</th>\n",
       "      <td>Herring</td>\n",
       "      <td>0.593750</td>\n",
       "      <td>0.037357</td>\n",
       "      <td>0.625000</td>\n",
       "      <td>0.531250</td>\n",
       "      <td>-0.062500</td>\n",
       "      <td>-0.093750</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>110</th>\n",
       "      <td>Symbols</td>\n",
       "      <td>0.780905</td>\n",
       "      <td>0.739564</td>\n",
       "      <td>0.806030</td>\n",
       "      <td>0.711558</td>\n",
       "      <td>-0.069347</td>\n",
       "      <td>-0.094472</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>CinCECGTorso</td>\n",
       "      <td>0.886957</td>\n",
       "      <td>0.663404</td>\n",
       "      <td>0.953623</td>\n",
       "      <td>0.802174</td>\n",
       "      <td>-0.084783</td>\n",
       "      <td>-0.151449</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>94</th>\n",
       "      <td>RefrigerationDevices</td>\n",
       "      <td>0.482667</td>\n",
       "      <td>0.982491</td>\n",
       "      <td>0.453333</td>\n",
       "      <td>0.378667</td>\n",
       "      <td>-0.104000</td>\n",
       "      <td>-0.074667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>111</th>\n",
       "      <td>SyntheticControl</td>\n",
       "      <td>0.946667</td>\n",
       "      <td>0.491556</td>\n",
       "      <td>0.993333</td>\n",
       "      <td>0.836667</td>\n",
       "      <td>-0.110000</td>\n",
       "      <td>-0.156667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>113</th>\n",
       "      <td>ToeSegmentation2</td>\n",
       "      <td>0.776923</td>\n",
       "      <td>1.354889</td>\n",
       "      <td>0.730769</td>\n",
       "      <td>0.638462</td>\n",
       "      <td>-0.138462</td>\n",
       "      <td>-0.092308</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>BirdChicken</td>\n",
       "      <td>0.850000</td>\n",
       "      <td>0.727008</td>\n",
       "      <td>0.700000</td>\n",
       "      <td>0.700000</td>\n",
       "      <td>-0.150000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>31</th>\n",
       "      <td>ECGFiveDays</td>\n",
       "      <td>0.777003</td>\n",
       "      <td>0.470631</td>\n",
       "      <td>0.881533</td>\n",
       "      <td>0.594657</td>\n",
       "      <td>-0.182346</td>\n",
       "      <td>-0.286876</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>BeetleFly</td>\n",
       "      <td>0.800000</td>\n",
       "      <td>0.892985</td>\n",
       "      <td>0.750000</td>\n",
       "      <td>0.550000</td>\n",
       "      <td>-0.250000</td>\n",
       "      <td>-0.200000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>DiatomSizeReduction</td>\n",
       "      <td>0.300654</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>45</th>\n",
       "      <td>Fungi</td>\n",
       "      <td>0.080645</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>56</th>\n",
       "      <td>HandOutlines</td>\n",
       "      <td>0.900000</td>\n",
       "      <td>0.014105</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>62</th>\n",
       "      <td>InsectEPGSmallTrain</td>\n",
       "      <td>0.566265</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>95</th>\n",
       "      <td>Rock</td>\n",
       "      <td>0.700000</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>97</th>\n",
       "      <td>SemgHandGenderCh2</td>\n",
       "      <td>0.836667</td>\n",
       "      <td>224.349365</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>98</th>\n",
       "      <td>SemgHandMovementCh2</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>99</th>\n",
       "      <td>SemgHandSubjectCh2</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>100</th>\n",
       "      <td>ShakeGestureWiimoteZ</td>\n",
       "      <td>0.100000</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>101</th>\n",
       "      <td>ShapeletSim</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>102</th>\n",
       "      <td>ShapesAll</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>103</th>\n",
       "      <td>SmallKitchenAppliances</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>105</th>\n",
       "      <td>SonyAIBORobotSurface1</td>\n",
       "      <td>0.613977</td>\n",
       "      <td>0.305926</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>128 rows × 7 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                              Name    Random        AE Loss  RandomCat  \\\n",
       "69                            Meat  0.333333       0.005748   0.950000   \n",
       "108                     Strawberry  0.586486       0.010888   0.959459   \n",
       "5                        ArrowHead  0.548571       0.100859   0.748571   \n",
       "68                          Mallat  0.776972       0.021429   0.939446   \n",
       "73     MiddlePhalanxOutlineCorrect  0.570447       0.009673   0.800687   \n",
       "19                        CricketZ  0.320513       0.484863   0.535897   \n",
       "92   ProximalPhalanxOutlineCorrect  0.646048       0.008137   0.890034   \n",
       "15                          Coffee  0.857143       0.016528   1.000000   \n",
       "37                        FaceFour  0.647727       0.734557   0.840909   \n",
       "6                             Beef  0.466667       0.167091   0.800000   \n",
       "127                           Yoga  0.581333       0.027104   0.794000   \n",
       "87                          PigCVP  0.048077       4.197798   0.221154   \n",
       "51                        GunPoint  0.813333       0.074187   0.960000   \n",
       "0                            ACSF1  0.490000       0.144930   0.480000   \n",
       "115                     TwoLeadECG  0.542581       0.141020   0.821773   \n",
       "17                        CricketX  0.371795       0.486353   0.569231   \n",
       "42                           FordB  0.661728       0.770388   0.817284   \n",
       "86                  PigArtPressure  0.072115     129.167374   0.201923   \n",
       "67                      Lightning7  0.671233       0.484338   0.643836   \n",
       "1                            Adiac  0.135550       0.013636   0.767263   \n",
       "23     DistalPhalanxOutlineCorrect  0.612319       0.027200   0.782609   \n",
       "40                            Fish  0.548571       0.028790   0.902857   \n",
       "93               ProximalPhalanxTW  0.731707       0.007223   0.804878   \n",
       "96                      ScreenType  0.357333       0.456331   0.354667   \n",
       "114                          Trace  0.580000       0.111137   0.940000   \n",
       "81                         OSULeaf  0.376033       0.471557   0.615702   \n",
       "82        PhalangesOutlinesCorrect  0.618881       0.015968   0.794872   \n",
       "112               ToeSegmentation1  0.552632       0.991212   0.657895   \n",
       "66                      Lightning2  0.737705       0.490883   0.622951   \n",
       "124                   WordSynonyms  0.476489       0.199155   0.583072   \n",
       "..                             ...       ...            ...        ...   \n",
       "116                    TwoPatterns  0.960000       0.231215   0.999250   \n",
       "33             EOGHorizontalSignal  0.477901    4723.121094   0.464088   \n",
       "83                         Phoneme  0.121308       0.883969   0.109705   \n",
       "85               PigAirwayPressure  0.120192       2.420203   0.120192   \n",
       "76           MixedShapesSmallTrain  0.842887       0.086877   0.838763   \n",
       "32                 ElectricDevices  0.714304       0.561187   0.694852   \n",
       "75         MixedShapesRegularTrain  0.896495       0.052715   0.922474   \n",
       "59                     HouseTwenty  0.722689  649137.250000   0.873950   \n",
       "58                         Herring  0.593750       0.037357   0.625000   \n",
       "110                        Symbols  0.780905       0.739564   0.806030   \n",
       "14                    CinCECGTorso  0.886957       0.663404   0.953623   \n",
       "94            RefrigerationDevices  0.482667       0.982491   0.453333   \n",
       "111               SyntheticControl  0.946667       0.491556   0.993333   \n",
       "113               ToeSegmentation2  0.776923       1.354889   0.730769   \n",
       "8                      BirdChicken  0.850000       0.727008   0.700000   \n",
       "31                     ECGFiveDays  0.777003       0.470631   0.881533   \n",
       "7                        BeetleFly  0.800000       0.892985   0.750000   \n",
       "21             DiatomSizeReduction  0.300654            NaN        NaN   \n",
       "45                           Fungi  0.080645            NaN        NaN   \n",
       "56                    HandOutlines  0.900000       0.014105        NaN   \n",
       "62             InsectEPGSmallTrain  0.566265            NaN        NaN   \n",
       "95                            Rock  0.700000            NaN        NaN   \n",
       "97               SemgHandGenderCh2  0.836667     224.349365        NaN   \n",
       "98             SemgHandMovementCh2       NaN            NaN        NaN   \n",
       "99              SemgHandSubjectCh2       NaN            NaN        NaN   \n",
       "100           ShakeGestureWiimoteZ  0.100000            NaN        NaN   \n",
       "101                    ShapeletSim       NaN            NaN        NaN   \n",
       "102                      ShapesAll       NaN            NaN        NaN   \n",
       "103         SmallKitchenAppliances       NaN            NaN        NaN   \n",
       "105          SonyAIBORobotSurface1  0.613977       0.305926        NaN   \n",
       "\n",
       "       Concat     GainT     GainC  \n",
       "69   0.650000  0.316667 -0.300000  \n",
       "108  0.832432  0.245946 -0.127027  \n",
       "5    0.788571  0.240000  0.040000  \n",
       "68   0.973134  0.196162  0.033689  \n",
       "73   0.762887  0.192440 -0.037801  \n",
       "19   0.502564  0.182051 -0.033333  \n",
       "92   0.817869  0.171821 -0.072165  \n",
       "15   1.000000  0.142857  0.000000  \n",
       "37   0.784091  0.136364 -0.056818  \n",
       "6    0.600000  0.133333 -0.200000  \n",
       "127  0.710333  0.129000 -0.083667  \n",
       "87   0.173077  0.125000 -0.048077  \n",
       "51   0.926667  0.113333 -0.033333  \n",
       "0    0.600000  0.110000  0.120000  \n",
       "115  0.651449  0.108867 -0.170325  \n",
       "17   0.479487  0.107692 -0.089744  \n",
       "42   0.769136  0.107407 -0.048148  \n",
       "86   0.177885  0.105769 -0.024038  \n",
       "67   0.767123  0.095890  0.123288  \n",
       "1    0.227621  0.092072 -0.539642  \n",
       "23   0.699275  0.086957 -0.083333  \n",
       "40   0.634286  0.085714 -0.268571  \n",
       "93   0.814634  0.082927  0.009756  \n",
       "96   0.440000  0.082667  0.085333  \n",
       "114  0.660000  0.080000 -0.280000  \n",
       "81   0.446281  0.070248 -0.169421  \n",
       "82   0.688811  0.069930 -0.106061  \n",
       "112  0.618421  0.065790 -0.039474  \n",
       "66   0.803279  0.065574  0.180328  \n",
       "124  0.540752  0.064263 -0.042320  \n",
       "..        ...       ...       ...  \n",
       "116  0.924500 -0.035500 -0.074750  \n",
       "33   0.439227 -0.038674 -0.024862  \n",
       "83   0.078059 -0.043249 -0.031646  \n",
       "85   0.072115 -0.048077 -0.048077  \n",
       "76   0.792990 -0.049897 -0.045773  \n",
       "32   0.663468 -0.050837 -0.031384  \n",
       "75   0.840000 -0.056495 -0.082474  \n",
       "59   0.663866 -0.058824 -0.210084  \n",
       "58   0.531250 -0.062500 -0.093750  \n",
       "110  0.711558 -0.069347 -0.094472  \n",
       "14   0.802174 -0.084783 -0.151449  \n",
       "94   0.378667 -0.104000 -0.074667  \n",
       "111  0.836667 -0.110000 -0.156667  \n",
       "113  0.638462 -0.138462 -0.092308  \n",
       "8    0.700000 -0.150000  0.000000  \n",
       "31   0.594657 -0.182346 -0.286876  \n",
       "7    0.550000 -0.250000 -0.200000  \n",
       "21        NaN       NaN       NaN  \n",
       "45        NaN       NaN       NaN  \n",
       "56        NaN       NaN       NaN  \n",
       "62        NaN       NaN       NaN  \n",
       "95        NaN       NaN       NaN  \n",
       "97        NaN       NaN       NaN  \n",
       "98        NaN       NaN       NaN  \n",
       "99        NaN       NaN       NaN  \n",
       "100       NaN       NaN       NaN  \n",
       "101       NaN       NaN       NaN  \n",
       "102       NaN       NaN       NaN  \n",
       "103       NaN       NaN       NaN  \n",
       "105       NaN       NaN       NaN  \n",
       "\n",
       "[128 rows x 7 columns]"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rDF.sort_values(\"GainT\",ascending=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1245,
   "metadata": {},
   "outputs": [],
   "source": [
    "learnTransfer.freeze_to(1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1247,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "Total time: 00:07 <p><table style='width:300px; margin-bottom:10px'>\n",
       "  <tr>\n",
       "    <th>epoch</th>\n",
       "    <th>train_loss</th>\n",
       "    <th>valid_loss</th>\n",
       "    <th>accuracy</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>1</th>\n",
       "    <th>0.093543</th>\n",
       "    <th>0.272463</th>\n",
       "    <th>0.935111</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>2</th>\n",
       "    <th>0.082593</th>\n",
       "    <th>0.271288</th>\n",
       "    <th>0.935556</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>3</th>\n",
       "    <th>0.080639</th>\n",
       "    <th>0.269597</th>\n",
       "    <th>0.936000</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>4</th>\n",
       "    <th>0.080529</th>\n",
       "    <th>0.268113</th>\n",
       "    <th>0.936889</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>5</th>\n",
       "    <th>0.078305</th>\n",
       "    <th>0.268081</th>\n",
       "    <th>0.936222</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>6</th>\n",
       "    <th>0.080295</th>\n",
       "    <th>0.268183</th>\n",
       "    <th>0.936889</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>7</th>\n",
       "    <th>0.078981</th>\n",
       "    <th>0.268396</th>\n",
       "    <th>0.936667</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>8</th>\n",
       "    <th>0.079904</th>\n",
       "    <th>0.268938</th>\n",
       "    <th>0.936222</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>9</th>\n",
       "    <th>0.078470</th>\n",
       "    <th>0.270798</th>\n",
       "    <th>0.936222</th>\n",
       "  </tr>\n",
       "  <tr>\n",
       "    <th>10</th>\n",
       "    <th>0.076618</th>\n",
       "    <th>0.270624</th>\n",
       "    <th>0.936444</th>\n",
       "  </tr>\n",
       "</table>\n"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "learnTransfer.fit_one_cycle(10,1e-3,wd=0.2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "def conv1d(ni:int, nf:int, ks:int=3, stride:int=1, padding:int=None, bias=False, init:LayerFunc=nn.init.kaiming_normal_) -> nn.Conv1d:\n",
    "    \"Create and initialize `nn.Conv1d` layer. `padding` defaults to `ks//2`.\"\n",
    "    if padding is None: padding = ks//2\n",
    "    return init_default(nn.Conv1d(ni, nf, kernel_size=ks, stride=stride, padding=padding, bias=bias), init)\n",
    "\n",
    "def _bn1d(ni, init_zero=False):\n",
    "    \"Batchnorm layer with 0 initialization\"\n",
    "    m = nn.BatchNorm1d(ni)\n",
    "    m.weight.data.fill_(0 if init_zero else 1)\n",
    "    m.bias.data.zero_()\n",
    "    return m\n",
    "\n",
    "def bn_relu_conv1d(ni, nf, ks, stride, init_zero=False):\n",
    "    bn_initzero = _bn1d(ni, init_zero=init_zero)\n",
    "    return nn.Sequential(bn_initzero, nn.ReLU(inplace=True), conv1d(ni, nf, ks, stride))\n",
    "\n",
    "class ResBlock(torch.nn.Module):\n",
    "    def __init__(self, ni, nf, stride, drop_p=0.0):\n",
    "        super().__init__()\n",
    "        self.bn = nn.BatchNorm1d(ni)\n",
    "        self.conv1 = conv1d(ni, nf, 3, stride)\n",
    "        self.conv2 = bn_relu_conv1d(nf, nf, 3, 1)\n",
    "        self.drop = nn.Dropout(drop_p, inplace=True) if drop_p else None\n",
    "        self.shortcut = conv1d(ni, nf, 1, stride) if ni != nf else noop\n",
    "\n",
    "    def forward(self, x):\n",
    "        x2 = F.relu(self.bn(x), inplace=True)\n",
    "        r = self.shortcut(x2)\n",
    "        x = self.conv1(x2)\n",
    "        if self.drop: x = self.drop(x)\n",
    "        x = self.conv2(x) * 0.2\n",
    "        return x.add_(r)\n",
    "\n",
    "def _make_group(N, ni, nf, block, stride, drop_p):\n",
    "    return [block(ni if i == 0 else nf, nf, stride if i == 0 else 1, drop_p) for i in range(N)]\n",
    "\n",
    "class WideResNet1d(nn.Module):\n",
    "    \"Wide ResNet with `num_groups` and a width of `k`.\"\n",
    "    def __init__(self, num_groups:int, N:int, k:int=1, drop_p:float=0.0, start_nf:int=16,maxY=20.0):\n",
    "        super().__init__()\n",
    "        self.maxY = maxY\n",
    "        n_channels = [start_nf]\n",
    "        for i in range(num_groups): n_channels.append(start_nf*(2**i)*k)\n",
    "        \n",
    "        \n",
    "        layers = [conv1d(1, n_channels[0], 3, 1)]  # conv1\n",
    "        self.split_groups = [layers[-1]] \n",
    "        \n",
    "        for i in range(num_groups):\n",
    "            layers += _make_group(N, n_channels[i], n_channels[i+1], ResBlock, (1 if i==0 else 2), drop_p)\n",
    "            self.split_groups.append(layers[-N])\n",
    "            \n",
    "        layers += [nn.BatchNorm1d(n_channels[-1]), nn.ReLU(inplace=True), nn.AdaptiveAvgPool1d(1),\n",
    "                   Flatten()]\n",
    "        #self.split_groups.append(layers[-1])\n",
    "        self.nf = n_channels[-1]\n",
    "        self.features = nn.Sequential(*layers)\n",
    "\n",
    "    def forward(self, x):\n",
    "        #x = x.unsqueeze(1)\n",
    "        return self.features(x)"
   ]
  },
  {
   "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": "3.6.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}