2017-10-25_FR_NN.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from torch import nn\n",
    "from tqdm import tqdm_notebook as tqdm\n",
    "import numpy as np\n",
    "import ipywidgets as widgets\n",
    "from ipywidgets import interact\n",
    "import pdb\n",
    "import pickle\n",
    "import time\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn.apionly as sns\n",
    "\n",
    "plt.style.use('clean')\n",
    "\n",
    "%run utils.py\n",
    "\n",
    "%matplotlib inline\n",
    "%config InlineBackend.figure_format = 'retina'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "w = widgets.IntText()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "04005be956624fe493bf68b6f747ac93",
       "version_major": 2,
       "version_minor": 0
      },
      "text/html": [
       "<p>Failed to display Jupyter Widget of type <code>interactive</code>.</p>\n",
       "<p>\n",
       "  If you're reading this message in the Jupyter Notebook or JupyterLab Notebook, it may mean\n",
       "  that the widgets JavaScript is still loading. If this message persists, it\n",
       "  likely means that the widgets JavaScript library is either not installed or\n",
       "  not enabled. See the <a href=\"https://ipywidgets.readthedocs.io/en/stable/user_install.html\">Jupyter\n",
       "  Widgets Documentation</a> for setup instructions.\n",
       "</p>\n",
       "<p>\n",
       "  If you're reading this message in another frontend (for example, a static\n",
       "  rendering on GitHub or <a href=\"https://nbviewer.jupyter.org/\">NBViewer</a>),\n",
       "  it may mean that your frontend doesn't currently support widgets.\n",
       "</p>\n"
      ],
      "text/plain": [
       "interactive(children=(IntText(value=0, description='x'), Output()), _dom_classes=('widget-interact',))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "@interact(x=w)\n",
    "def f(x):\n",
    "    WINDOW_SIZE = 100\n",
    "    \n",
    "    epoch_progress = len(losses[-1])\n",
    "    i = max(epoch_progress-WINDOW_SIZE,0)\n",
    "    if len(losses) > 1:\n",
    "        diffs = np.array(losses[-1][i:epoch_progress]) - np.array(losses[-2][i:epoch_progress])\n",
    "        plt.plot(range(i,epoch_progress), diffs, label='This-Last Epoch')\n",
    "    else:\n",
    "        plt.plot(range(i,epoch_progress), losses[-1][i:epoch_progress], label='This Epoch')\n",
    "    plt.xlim((i,epoch_progress))\n",
    "    plt.legend()\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def df_to_torch(df_clean):\n",
    "    y = Variable(torch.from_numpy(df_clean['duration'].as_matrix()).float())\n",
    "    x = Variable(torch.from_numpy(df_clean.drop(['duration',], axis=1).as_matrix()).float())\n",
    "    return x, y"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "D_in, H, D_out = 36, 50, 1\n",
    "model = nn.Sequential(\n",
    "    nn.Linear(D_in, H, bias=False),\n",
    "    nn.Tanh(),\n",
    "    nn.Linear(H, D_out, bias=False),\n",
    ")\n",
    "\n",
    "loss_fn = torch.nn.MSELoss(size_average=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "losses = []"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "97cfe4cf044a46dfa1d31d71a3617b11",
       "version_major": 2,
       "version_minor": 0
      },
      "text/html": [
       "<p>Failed to display Jupyter Widget of type <code>HBox</code>.</p>\n",
       "<p>\n",
       "  If you're reading this message in the Jupyter Notebook or JupyterLab Notebook, it may mean\n",
       "  that the widgets JavaScript is still loading. If this message persists, it\n",
       "  likely means that the widgets JavaScript library is either not installed or\n",
       "  not enabled. See the <a href=\"https://ipywidgets.readthedocs.io/en/stable/user_install.html\">Jupyter\n",
       "  Widgets Documentation</a> for setup instructions.\n",
       "</p>\n",
       "<p>\n",
       "  If you're reading this message in another frontend (for example, a static\n",
       "  rendering on GitHub or <a href=\"https://nbviewer.jupyter.org/\">NBViewer</a>),\n",
       "  it may mean that your frontend doesn't currently support widgets.\n",
       "</p>\n"
      ],
      "text/plain": [
       "HBox(children=(IntProgress(value=0, max=404525760), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 0 loss 55256.13952797469\n",
      "epoch 1 loss 43497.16038469342\n",
      "epoch 2 loss 42561.275177036936\n",
      "epoch 3 loss 42100.353360142006\n",
      "epoch 4 loss 41861.67541369924\n",
      "epoch 5 loss 41711.43039938915\n",
      "epoch 6 loss 41587.58053296779\n",
      "epoch 7 loss 41456.355246943975\n",
      "epoch 8 loss 41348.80136726027\n",
      "epoch 9 loss 41279.321745492794\n",
      "epoch 10 loss 41219.634572987394\n",
      "epoch 11 loss 41187.15896718172\n",
      "epoch 12 loss 41145.81866633166\n",
      "\n"
     ]
    },
    {
     "ename": "KeyboardInterrupt",
     "evalue": "",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-7-4fd4b5869d83>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m     12\u001b[0m         \u001b[0mdf_iter\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mpd\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mread_csv\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m'data/yellow_tripdata_2009-01_train.csv'\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mchunksize\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mchunksize\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     13\u001b[0m         \u001b[0mtotal_loss\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;36m0\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 14\u001b[0;31m         \u001b[0;32mfor\u001b[0m \u001b[0mdf\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mdf_iter\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     15\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     16\u001b[0m             \u001b[0mdf\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mdf\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mloc\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdf\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m'duration'\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;34m<\u001b[0m \u001b[0;36m3600\u001b[0m \u001b[0;34m*\u001b[0m \u001b[0;36m2\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;34m&\u001b[0m \u001b[0;34m(\u001b[0m\u001b[0mdf\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m'duration'\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;34m>\u001b[0m \u001b[0;36m120\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/anaconda/envs/py36/lib/python3.6/site-packages/pandas/io/parsers.py\u001b[0m in \u001b[0;36m__next__\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m    976\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0m__next__\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    977\u001b[0m         \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 978\u001b[0;31m             \u001b[0;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mget_chunk\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    979\u001b[0m         \u001b[0;32mexcept\u001b[0m \u001b[0mStopIteration\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    980\u001b[0m             \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mclose\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/anaconda/envs/py36/lib/python3.6/site-packages/pandas/io/parsers.py\u001b[0m in \u001b[0;36mget_chunk\u001b[0;34m(self, size)\u001b[0m\n\u001b[1;32m   1040\u001b[0m                 \u001b[0;32mraise\u001b[0m \u001b[0mStopIteration\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1041\u001b[0m             \u001b[0msize\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mmin\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0msize\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mnrows\u001b[0m \u001b[0;34m-\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_currow\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1042\u001b[0;31m         \u001b[0;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mread\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mnrows\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0msize\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1043\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1044\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/anaconda/envs/py36/lib/python3.6/site-packages/pandas/io/parsers.py\u001b[0m in \u001b[0;36mread\u001b[0;34m(self, nrows)\u001b[0m\n\u001b[1;32m   1003\u001b[0m                 \u001b[0;32mraise\u001b[0m \u001b[0mValueError\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m'skipfooter not supported for iteration'\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1004\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1005\u001b[0;31m         \u001b[0mret\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_engine\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mread\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mnrows\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1006\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1007\u001b[0m         \u001b[0;32mif\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0moptions\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mget\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m'as_recarray'\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/anaconda/envs/py36/lib/python3.6/site-packages/pandas/io/parsers.py\u001b[0m in \u001b[0;36mread\u001b[0;34m(self, nrows)\u001b[0m\n\u001b[1;32m   1746\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0mread\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mnrows\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mNone\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1747\u001b[0m         \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1748\u001b[0;31m             \u001b[0mdata\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_reader\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mread\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mnrows\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1749\u001b[0m         \u001b[0;32mexcept\u001b[0m \u001b[0mStopIteration\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1750\u001b[0m             \u001b[0;32mif\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_first_chunk\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32mpandas/_libs/parsers.pyx\u001b[0m in \u001b[0;36mpandas._libs.parsers.TextReader.read (pandas/_libs/parsers.c:10862)\u001b[0;34m()\u001b[0m\n",
      "\u001b[0;32mpandas/_libs/parsers.pyx\u001b[0m in \u001b[0;36mpandas._libs.parsers.TextReader._read_low_memory (pandas/_libs/parsers.c:11343)\u001b[0;34m()\u001b[0m\n",
      "\u001b[0;32mpandas/_libs/parsers.pyx\u001b[0m in \u001b[0;36mpandas._libs.parsers.TextReader._read_rows (pandas/_libs/parsers.c:12175)\u001b[0;34m()\u001b[0m\n",
      "\u001b[0;32mpandas/_libs/parsers.pyx\u001b[0m in \u001b[0;36mpandas._libs.parsers.TextReader._convert_column_data (pandas/_libs/parsers.c:14136)\u001b[0;34m()\u001b[0m\n",
      "\u001b[0;32mpandas/_libs/parsers.pyx\u001b[0m in \u001b[0;36mpandas._libs.parsers.TextReader._convert_tokens (pandas/_libs/parsers.c:14858)\u001b[0;34m()\u001b[0m\n",
      "\u001b[0;32mpandas/_libs/parsers.pyx\u001b[0m in \u001b[0;36mpandas._libs.parsers.TextReader._convert_with_dtype (pandas/_libs/parsers.c:15629)\u001b[0;34m()\u001b[0m\n",
      "\u001b[0;32m/anaconda/envs/py36/lib/python3.6/site-packages/pandas/core/dtypes/common.py\u001b[0m in \u001b[0;36mis_integer_dtype\u001b[0;34m(arr_or_dtype)\u001b[0m\n\u001b[1;32m    738\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    739\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 740\u001b[0;31m \u001b[0;32mdef\u001b[0m \u001b[0mis_integer_dtype\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0marr_or_dtype\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    741\u001b[0m     \"\"\"\n\u001b[1;32m    742\u001b[0m     \u001b[0mCheck\u001b[0m \u001b[0mwhether\u001b[0m \u001b[0mthe\u001b[0m \u001b[0mprovided\u001b[0m \u001b[0marray\u001b[0m \u001b[0;32mor\u001b[0m \u001b[0mdtype\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0mof\u001b[0m \u001b[0man\u001b[0m \u001b[0minteger\u001b[0m \u001b[0mdtype\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m: "
     ]
    }
   ],
   "source": [
    "i = 0\n",
    "num_entries = 13484192\n",
    "chunksize = 10000\n",
    "w.value = 0\n",
    "learning_rate = 1e-7\n",
    "display_freq = 10\n",
    "\n",
    "num_epochs = 30\n",
    "with tqdm(total=num_epochs*num_entries) as pbar:\n",
    "    for epoch in range(num_epochs):\n",
    "        losses.append([])\n",
    "        df_iter = pd.read_csv('data/yellow_tripdata_2009-01_train.csv', chunksize=chunksize)\n",
    "        total_loss = 0\n",
    "        for df in df_iter:\n",
    "            \n",
    "            df = df.loc[(df['duration'] < 3600 * 2) & (df['duration'] > 120),:]\n",
    "            x, y = df_to_torch(df)\n",
    "            pbar.update(1)\n",
    "            # Forward pass: compute predicted y by passing x to the model. Module objects\n",
    "            # override the _call_ operator so you can call them like functions. When\n",
    "            # doing so you pass a Variable of input data to the Module and it produces\n",
    "            # a Variable of output data.\n",
    "            y_pred = model(x)\n",
    "\n",
    "            # Compute and print loss. We pass Variables containing the predicted and true\n",
    "            # values of y, and the loss function returns a Variable containing the loss.\n",
    "            loss = loss_fn(y_pred, y)\n",
    "            total_loss += loss.data[0]\n",
    "            \n",
    "            # dashboard logic\n",
    "            losses[-1].append(loss.data[0]/chunksize)\n",
    "            i += 1\n",
    "            if (i % display_freq == 0):\n",
    "                w.value = w.value + 1\n",
    "            \n",
    "            # Zero the gradients before running the backward pass.\n",
    "            model.zero_grad()\n",
    "\n",
    "            # Backward pass: compute gradient of the loss with respect to all the learnable\n",
    "            # parameters of the model. Internally, the parameters of each Module are stored\n",
    "            # in Variables with requires_grad=True, so this call will compute gradients for\n",
    "            # all learnable parameters in the model.\n",
    "            loss.backward()\n",
    "\n",
    "            # Update the weights using gradient descent. Each parameter is a Variable, so\n",
    "            # we can access its data and gradients like we did before.\n",
    "            for param in model.parameters():\n",
    "                param.data -= learning_rate * param.grad.data\n",
    "                \n",
    "        print('epoch', epoch, 'loss', total_loss/num_entries)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "ename": "AttributeError",
     "evalue": "'Sequential' object has no attribute 'save_state_dict'",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mAttributeError\u001b[0m                            Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-8-2b48a6d22ba3>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m \u001b[0mmodel\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msave_state_dict\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m'results/FR_nn_weights.pt'\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
      "\u001b[0;32m/anaconda/envs/py36/lib/python3.6/site-packages/torch/nn/modules/module.py\u001b[0m in \u001b[0;36m__getattr__\u001b[0;34m(self, name)\u001b[0m\n\u001b[1;32m    260\u001b[0m                 \u001b[0;32mreturn\u001b[0m \u001b[0mmodules\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mname\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    261\u001b[0m         raise AttributeError(\"'{}' object has no attribute '{}'\".format(\n\u001b[0;32m--> 262\u001b[0;31m             type(self).__name__, name))\n\u001b[0m\u001b[1;32m    263\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    264\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0m__setattr__\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mname\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mvalue\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mAttributeError\u001b[0m: 'Sequential' object has no attribute 'save_state_dict'"
     ]
    }
   ],
   "source": [
    "model.save_state_dict('results/FR_nn_weights.pt')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "d = model.state_dict()\n",
    "pickle.dump(d, open('results/FR_nn_weights.pt', 'wb'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [],
   "source": [
    "test_df = pd.read_csv('data/yellow_tripdata_2009-01_test.csv')\n",
    "test_df = test_df.loc[(test_df['duration'] < 3600 * 2) & (test_df['duration'] > 120),:]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.6797969394014256"
      ]
     },
     "execution_count": 46,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "1 - np.std(60*errs)**2 / test_df.duration.std()**2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "collapsed": true
   },
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "20aeda1d5da24b7c9eb04aa925d72063",
       "version_major": 2,
       "version_minor": 0
      },
      "text/html": [
       "<p>Failed to display Jupyter Widget of type <code>HBox</code>.</p>\n",
       "<p>\n",
       "  If you're reading this message in the Jupyter Notebook or JupyterLab Notebook, it may mean\n",
       "  that the widgets JavaScript is still loading. If this message persists, it\n",
       "  likely means that the widgets JavaScript library is either not installed or\n",
       "  not enabled. See the <a href=\"https://ipywidgets.readthedocs.io/en/stable/user_install.html\">Jupyter\n",
       "  Widgets Documentation</a> for setup instructions.\n",
       "</p>\n",
       "<p>\n",
       "  If you're reading this message in another frontend (for example, a static\n",
       "  rendering on GitHub or <a href=\"https://nbviewer.jupyter.org/\">NBViewer</a>),\n",
       "  it may mean that your frontend doesn't currently support widgets.\n",
       "</p>\n"
      ],
      "text/plain": [
       "HBox(children=(IntProgress(value=1, bar_style='info', max=1), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "errs = []\n",
    "test_iter = pd.read_csv('data/yellow_tripdata_2009-01_test.csv', chunksize=100000)\n",
    "for df in tqdm(test_iter):\n",
    "    df = df.loc[(df['duration'] < 3600 * 2) & (df['duration'] > 120),:]\n",
    "    x, y = df_to_torch(df)\n",
    "    errs += list((y - model(x).squeeze()).data.numpy()/60)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['duration', 'Trip_Distance', 'Start_Lon', 'Start_Lat', 'End_Lon',\n",
       "       'End_Lat', 'week_0', 'week_1', 'week_2', 'week_3', 'week_4', 'week_5',\n",
       "       'week_6', 'hour_0', 'hour_1', 'hour_2', 'hour_3', 'hour_4', 'hour_5',\n",
       "       'hour_6', 'hour_7', 'hour_8', 'hour_9', 'hour_10', 'hour_11', 'hour_12',\n",
       "       'hour_13', 'hour_14', 'hour_15', 'hour_16', 'hour_17', 'hour_18',\n",
       "       'hour_19', 'hour_20', 'hour_21', 'hour_22', 'hour_23'],\n",
       "      dtype='object')"
      ]
     },
     "execution_count": 47,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.columns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_filtered_errs(day, hour):\n",
    "    errs = []\n",
    "    test_iter = pd.read_csv('data/yellow_tripdata_2009-01_test.csv', chunksize=100000)\n",
    "    for df in tqdm(test_iter):\n",
    "        df = df.loc[(df['duration'] < 3600 * 2) & (df['duration'] > 120) &\n",
    "                    (df['week_{}'.format(day)] == 1) & (df['hour_{}'.format(hour)] == 1),:]\n",
    "        x, y = df_to_torch(df)\n",
    "        errs += list((y - model(x).squeeze()).data.numpy()/60)\n",
    "    return np.array(errs)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "0a9b1ce06041431e996232df8c6f3dbc",
       "version_major": 2,
       "version_minor": 0
      },
      "text/html": [
       "<p>Failed to display Jupyter Widget of type <code>HBox</code>.</p>\n",
       "<p>\n",
       "  If you're reading this message in the Jupyter Notebook or JupyterLab Notebook, it may mean\n",
       "  that the widgets JavaScript is still loading. If this message persists, it\n",
       "  likely means that the widgets JavaScript library is either not installed or\n",
       "  not enabled. See the <a href=\"https://ipywidgets.readthedocs.io/en/stable/user_install.html\">Jupyter\n",
       "  Widgets Documentation</a> for setup instructions.\n",
       "</p>\n",
       "<p>\n",
       "  If you're reading this message in another frontend (for example, a static\n",
       "  rendering on GitHub or <a href=\"https://nbviewer.jupyter.org/\">NBViewer</a>),\n",
       "  it may mean that your frontend doesn't currently support widgets.\n",
       "</p>\n"
      ],
      "text/plain": [
       "HBox(children=(IntProgress(value=1, bar_style='info', max=1), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "tues0errs = get_filtered_errs(1, 0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "OrderedDict([('0.weight', \n",
       "               -6.7131  -5.3898  -5.2211  ...   -0.5684  -0.5240  -0.3523\n",
       "               -0.1239   0.3352   0.5233  ...   -1.8631  -1.8280  -2.0494\n",
       "               -0.7041   0.1179  -0.0568  ...   -5.5683  -5.8238  -5.9347\n",
       "                         ...               ⋱              ...            \n",
       "                3.4464  -1.7407  -1.9854  ...    3.1583   3.4001   3.0320\n",
       "               -0.4470   0.0892   0.1278  ...   -0.6920   0.5706  -1.6982\n",
       "               -6.6336   3.4225   1.5398  ...    0.7933   1.2883   1.3808\n",
       "              [torch.FloatTensor of size 50x36]), ('2.weight', \n",
       "              \n",
       "              Columns 0 to 7 \n",
       "              -38.7345 -45.9827  50.5846  57.9655 -38.2473  35.0266 -42.3580 -39.7950\n",
       "              \n",
       "              Columns 8 to 15 \n",
       "              -46.4611  43.8492  48.7893 -60.1340 -49.5148  46.4025 -70.2076 -46.0062\n",
       "              \n",
       "              Columns 16 to 23 \n",
       "               48.1139 -34.1263  45.4359 -46.3013  35.8642 -53.1262  85.7021  41.6838\n",
       "              \n",
       "              Columns 24 to 31 \n",
       "               37.3086  48.8178 -32.5046 -40.9912  46.1815 -72.9474  42.4307 -38.9761\n",
       "              \n",
       "              Columns 32 to 39 \n",
       "              -44.7674  50.0544 -79.3992 -49.2383 -45.4019  38.2222  37.6180 -33.4122\n",
       "              \n",
       "              Columns 40 to 47 \n",
       "              -44.5154  42.1094  48.4169 -40.7151  32.9308  39.5189 -41.5034  32.5961\n",
       "              \n",
       "              Columns 48 to 49 \n",
       "              -43.2765 -43.6276\n",
       "              [torch.FloatTensor of size 1x50])])"
      ]
     },
     "execution_count": 64,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.state_dict()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "3.7181666"
      ]
     },
     "execution_count": 55,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.std(tues0errs)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "errs = np.array(errs)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "collapsed": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "-0.073984928"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.mean(errs)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "3.0173934"
      ]
     },
     "execution_count": 60,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.std(np.abs(errs))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "6.853783251669408"
      ]
     },
     "execution_count": 59,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test_df.duration.std() / 60"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "3.8783154"
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "12.790218696594224"
      ]
     },
     "execution_count": 62,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.percentile(np.abs(errs), 99)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {
    "collapsed": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAY0AAAD8CAYAAACLrvgBAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x11b26ff98>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.hist(errs, bins=np.arange(-20,20.1,.5));"
   ]
  },
  {
   "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>duration</th>\n",
       "      <th>Trip_Distance</th>\n",
       "      <th>Start_Lon</th>\n",
       "      <th>Start_Lat</th>\n",
       "      <th>End_Lon</th>\n",
       "      <th>End_Lat</th>\n",
       "      <th>week_0</th>\n",
       "      <th>week_1</th>\n",
       "      <th>week_2</th>\n",
       "      <th>week_3</th>\n",
       "      <th>...</th>\n",
       "      <th>hour_14</th>\n",
       "      <th>hour_15</th>\n",
       "      <th>hour_16</th>\n",
       "      <th>hour_17</th>\n",
       "      <th>hour_18</th>\n",
       "      <th>hour_19</th>\n",
       "      <th>hour_20</th>\n",
       "      <th>hour_21</th>\n",
       "      <th>hour_22</th>\n",
       "      <th>hour_23</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1300000</th>\n",
       "      <td>306.0</td>\n",
       "      <td>-0.684758</td>\n",
       "      <td>0.549058</td>\n",
       "      <td>0.956021</td>\n",
       "      <td>0.304054</td>\n",
       "      <td>0.462084</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>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",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1300001</th>\n",
       "      <td>864.0</td>\n",
       "      <td>1.459473</td>\n",
       "      <td>1.020434</td>\n",
       "      <td>0.971887</td>\n",
       "      <td>-0.520599</td>\n",
       "      <td>-1.129734</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>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1300002</th>\n",
       "      <td>192.0</td>\n",
       "      <td>-0.593514</td>\n",
       "      <td>-0.508017</td>\n",
       "      <td>0.162640</td>\n",
       "      <td>-0.113236</td>\n",
       "      <td>0.457010</td>\n",
       "      <td>0</td>\n",
       "      <td>1</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>1300003</th>\n",
       "      <td>136.0</td>\n",
       "      <td>-0.776002</td>\n",
       "      <td>-0.268128</td>\n",
       "      <td>0.706322</td>\n",
       "      <td>-0.388813</td>\n",
       "      <td>0.360499</td>\n",
       "      <td>0</td>\n",
       "      <td>1</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>1300004</th>\n",
       "      <td>376.0</td>\n",
       "      <td>-0.411026</td>\n",
       "      <td>-0.123961</td>\n",
       "      <td>1.337730</td>\n",
       "      <td>0.258689</td>\n",
       "      <td>1.802127</td>\n",
       "      <td>0</td>\n",
       "      <td>1</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>1</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 × 37 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "         duration  Trip_Distance  Start_Lon  Start_Lat   End_Lon   End_Lat  \\\n",
       "1300000     306.0      -0.684758   0.549058   0.956021  0.304054  0.462084   \n",
       "1300001     864.0       1.459473   1.020434   0.971887 -0.520599 -1.129734   \n",
       "1300002     192.0      -0.593514  -0.508017   0.162640 -0.113236  0.457010   \n",
       "1300003     136.0      -0.776002  -0.268128   0.706322 -0.388813  0.360499   \n",
       "1300004     376.0      -0.411026  -0.123961   1.337730  0.258689  1.802127   \n",
       "\n",
       "         week_0  week_1  week_2  week_3   ...     hour_14  hour_15  hour_16  \\\n",
       "1300000       1       0       0       0   ...           0        0        0   \n",
       "1300001       0       0       0       0   ...           0        0        0   \n",
       "1300002       0       1       0       0   ...           0        0        0   \n",
       "1300003       0       1       0       0   ...           0        0        0   \n",
       "1300004       0       1       0       0   ...           0        0        0   \n",
       "\n",
       "         hour_17  hour_18  hour_19  hour_20  hour_21  hour_22  hour_23  \n",
       "1300000        0        1        0        0        0        0        0  \n",
       "1300001        0        0        0        1        0        0        0  \n",
       "1300002        0        0        0        0        0        0        0  \n",
       "1300003        0        0        0        0        0        0        0  \n",
       "1300004        1        0        0        0        0        0        0  \n",
       "\n",
       "[5 rows x 37 columns]"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.text.Text at 0x1192dbfd0>"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAwgAAAIOCAYAAAAcKyq+AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nJEnqGSBIkiRJ6v1/dep0uE4STq4AAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x118060748>"
      ]
     },
     "metadata": {
      "image/png": {
       "height": 263,
       "width": 388
      }
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "y = np.array([.54, .68, .33, .53, .53])\n",
    "models = ['GMaps', 'NN', 'Lin Reg', 'Uniform', 'Softmax']\n",
    "x = np.arange(len(y))\n",
    "\n",
    "\n",
    "plt.bar(x, y)\n",
    "plt.xticks(x, models)\n",
    "for x1, y1 in zip(x,y):\n",
    "    plt.annotate(str(y1), (x1,y1+.01), ha='center')\n",
    "sns.despine()\n",
    "\n",
    "plt.title('Comparison of Models')\n",
    "plt.ylabel('Out of Sample $R^2$')"
   ]
  },
  {
   "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": "3.6.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}