Demo.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Boat net\n",
    "\n",
    "Dashboard: http://localhost:9080"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "import random\n",
    "import collections\n",
    "import itertools\n",
    "import numpy as np\n",
    "from sklearn.metrics import confusion_matrix\n",
    "import matplotlib.pyplot as plt\n",
    "import torch\n",
    "from torch.utils.data import DataLoader\n",
    "from torch.autograd import Variable\n",
    "import demo\n",
    "from demo import BoatDataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "128"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dataset = BoatDataset(\"data.csv\", \"sounds-spectrogram.pkl\")\n",
    "dataloader = DataLoader(\n",
    "    dataset,\n",
    "    batch_size=1,\n",
    "    num_workers=4,\n",
    "    shuffle=True\n",
    ")\n",
    "input_size = dataset[0][0].size(1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "net = demo.BoatNet(input_size=input_size)\n",
    "net.train(False)  # set dropout to inference mode\n",
    "state_dict = torch.load(\"model_00_000001999.state\")\n",
    "net.load_state_dict(state_dict=state_dict)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "x, y = dataset[-1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "def pred(x):\n",
    "    x = x.expand(1, -1, -1)  # expand batch of one\n",
    "    # Reset hidden layer!\n",
    "    hidden = net.init_hidden()\n",
    "    hidden.volatile=True\n",
    "    return net(Variable(x, volatile=True), hidden).data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['-0.000', '-9.528', '-11.165']\n",
      "0\n",
      "0\n",
      "564537\n"
     ]
    }
   ],
   "source": [
    "y_pred = pred(x)\n",
    "print [\"%.3f\" % i for i in y_pred.view(-1).tolist()]\n",
    "print np.array(pred(x)).argmax()\n",
    "print y\n",
    "print len(dataset)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "got 907 of 1000: %90.70\n",
      "saw Counter({2: 513, 1: 365, 0: 122})\n"
     ]
    }
   ],
   "source": [
    "indices = range(len(dataset))\n",
    "random.Random(0).shuffle(indices)\n",
    "right = []\n",
    "labels = []\n",
    "labels_pred = []\n",
    "for i in indices[:1000]:\n",
    "    x, y = dataset[i]\n",
    "    y_pred = pred(x)\n",
    "    guess = pred(x).numpy().argmax()\n",
    "    right.append(guess == y)\n",
    "    labels.append(y)\n",
    "    labels_pred.append(guess)\n",
    "\n",
    "l = len(right)\n",
    "r = sum(right)\n",
    "print \"got %d of %d: %%%.2f\" % (r, l, 100. * r / l)\n",
    "print \"saw\", collections.Counter(labels)\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[114,   1,   7],\n",
       "       [  2, 303,  60],\n",
       "       [  3,  20, 490]])"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cnf_matrix = confusion_matrix(labels, labels_pred)\n",
    "cnf_matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0.5,-30.5079,u'Predicted label')"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f2e44d637d0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.imshow(cnf_matrix, interpolation='nearest', cmap=plt.cm.Blues)\n",
    "classes = [\"Neither\", \"T1 - Bottom\", \"T2 - Midwater\"]\n",
    "plt.colorbar()\n",
    "tick_marks = np.arange(len(classes))\n",
    "plt.xticks(tick_marks, classes, rotation=45)\n",
    "plt.yticks(tick_marks, classes)\n",
    "\n",
    "fmt = 'd'\n",
    "thresh = cnf_matrix.max() / 2.\n",
    "for i, j in itertools.product(range(cnf_matrix.shape[0]), range(cnf_matrix.shape[1])):\n",
    "    plt.text(j, i, format(cnf_matrix[i, j], fmt),\n",
    "             horizontalalignment=\"center\",\n",
    "             color=\"white\" if cnf_matrix[i, j] > thresh else \"black\")\n",
    "\n",
    "plt.tight_layout()\n",
    "plt.ylabel('True label')\n",
    "plt.xlabel('Predicted label')\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Environment (conda_arel-pytorch)",
   "language": "python",
   "name": "conda_arel-pytorch"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.14"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}