tabular-example.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "%reload_ext autoreload\n",
    "%autoreload 2\n",
    "%matplotlib inline\n",
    "\n",
    "from fastai import *\n",
    "from fastai.tabular import *\n",
    "from sklearn.metrics import confusion_matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "path = Path('data/ham10000')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "df = pickle.load(open(path/'df.pkl', 'rb'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "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>lesion_id</th>\n",
       "      <th>image_id</th>\n",
       "      <th>dx</th>\n",
       "      <th>dx_type</th>\n",
       "      <th>age</th>\n",
       "      <th>sex</th>\n",
       "      <th>localization</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>9687</th>\n",
       "      <td>HAM_0002644</td>\n",
       "      <td>ISIC_0029417</td>\n",
       "      <td>akiec</td>\n",
       "      <td>histo</td>\n",
       "      <td>80.0</td>\n",
       "      <td>female</td>\n",
       "      <td>neck</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9688</th>\n",
       "      <td>HAM_0006002</td>\n",
       "      <td>ISIC_0029915</td>\n",
       "      <td>akiec</td>\n",
       "      <td>histo</td>\n",
       "      <td>50.0</td>\n",
       "      <td>female</td>\n",
       "      <td>face</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9689</th>\n",
       "      <td>HAM_0000549</td>\n",
       "      <td>ISIC_0029360</td>\n",
       "      <td>akiec</td>\n",
       "      <td>histo</td>\n",
       "      <td>70.0</td>\n",
       "      <td>male</td>\n",
       "      <td>upper extremity</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9690</th>\n",
       "      <td>HAM_0000549</td>\n",
       "      <td>ISIC_0026152</td>\n",
       "      <td>akiec</td>\n",
       "      <td>histo</td>\n",
       "      <td>70.0</td>\n",
       "      <td>male</td>\n",
       "      <td>upper extremity</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9691</th>\n",
       "      <td>HAM_0000673</td>\n",
       "      <td>ISIC_0029659</td>\n",
       "      <td>akiec</td>\n",
       "      <td>histo</td>\n",
       "      <td>70.0</td>\n",
       "      <td>female</td>\n",
       "      <td>face</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        lesion_id      image_id     dx dx_type   age     sex     localization\n",
       "9687  HAM_0002644  ISIC_0029417  akiec   histo  80.0  female             neck\n",
       "9688  HAM_0006002  ISIC_0029915  akiec   histo  50.0  female             face\n",
       "9689  HAM_0000549  ISIC_0029360  akiec   histo  70.0    male  upper extremity\n",
       "9690  HAM_0000549  ISIC_0026152  akiec   histo  70.0    male  upper extremity\n",
       "9691  HAM_0000673  ISIC_0029659  akiec   histo  70.0  female             face"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "cat_names = ['sex', 'localization']\n",
    "cont_names = ['age']\n",
    "dep_var = 'dx'\n",
    "procs = [FillMissing, Categorify, Normalize]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "test = TabularList.from_df(df, path=path, cat_names=cat_names, cont_names=cont_names, procs=procs)\n",
    "data = LabelLists.load_empty(path).add_test(test).databunch()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "learn = tabular_learner(data, layers=[128,64], metrics=accuracy)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "learn.load('tabular');"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['akiec', 'bcc', 'bkl', 'df', 'mel', 'nv', 'vasc']"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data.valid_ds.y.classes"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Predict using learn.get_preds()\n",
    "\n",
    "Prediciton is happening on GPU on Test dataframe"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "probs, _ = learn.get_preds(DatasetType.Test)\n",
    "preds = probs.argmax(dim=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tensor([2, 2, 5, 5, 2, 2, 2, 5, 5, 2, 2, 5, 4, 2, 2, 5, 2, 2, 5, 2, 5, 2, 5, 5,\n",
       "        5, 5, 2, 2, 2, 2, 2, 5, 5, 5, 5, 5, 5, 2, 2, 5])"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "preds"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "p = ['akiec' for x in preds if x == 2]\n",
    "p += ['mel' for y in preds if y == 4]\n",
    "p += ['nv' for y in preds if y == 5]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[20,  0,  0,  0],\n",
       "       [ 0,  0,  1, 19],\n",
       "       [ 0,  0,  0,  0],\n",
       "       [ 0,  0,  0,  0]])"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "confusion_matrix(df.dx, p)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Predict using learn.predict()\n",
    "\n",
    "Prediciton is happening on CPU."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "preds = df.apply(learn.predict, axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "9687    (nv, tensor(5), [tensor(0.0001), tensor(0.0029...\n",
       "9688    (nv, tensor(5), [tensor(0.0001), tensor(0.0026...\n",
       "9689    (nv, tensor(5), [tensor(9.1095e-05), tensor(0....\n",
       "9690    (nv, tensor(5), [tensor(9.1095e-05), tensor(0....\n",
       "9691    (nv, tensor(5), [tensor(0.0001), tensor(0.0028...\n",
       "9692    (nv, tensor(5), [tensor(0.0003), tensor(0.0039...\n",
       "9693    (nv, tensor(5), [tensor(0.0003), tensor(0.0039...\n",
       "9694    (nv, tensor(5), [tensor(5.6556e-05), tensor(0....\n",
       "9695    (nv, tensor(5), [tensor(5.6556e-05), tensor(0....\n",
       "9696    (nv, tensor(5), [tensor(0.0002), tensor(0.0025...\n",
       "9697    (nv, tensor(5), [tensor(0.0002), tensor(0.0025...\n",
       "9698    (nv, tensor(5), [tensor(5.6556e-05), tensor(0....\n",
       "9699    (nv, tensor(5), [tensor(6.7162e-05), tensor(0....\n",
       "9700    (nv, tensor(5), [tensor(0.0002), tensor(0.0025...\n",
       "9701    (nv, tensor(5), [tensor(0.0002), tensor(0.0025...\n",
       "9702    (nv, tensor(5), [tensor(9.5781e-05), tensor(0....\n",
       "9703    (nv, tensor(5), [tensor(0.0002), tensor(0.0051...\n",
       "9704    (nv, tensor(5), [tensor(0.0001), tensor(0.0046...\n",
       "9705    (nv, tensor(5), [tensor(0.0001), tensor(0.0027...\n",
       "9706    (nv, tensor(5), [tensor(6.3124e-05), tensor(0....\n",
       "2462    (nv, tensor(5), [tensor(0.0002), tensor(0.0045...\n",
       "2463    (nv, tensor(5), [tensor(0.0001), tensor(0.0028...\n",
       "2464    (nv, tensor(5), [tensor(5.6529e-05), tensor(0....\n",
       "2465    (nv, tensor(5), [tensor(5.6529e-05), tensor(0....\n",
       "2466    (nv, tensor(5), [tensor(0.0002), tensor(0.0045...\n",
       "2467    (nv, tensor(5), [tensor(0.0002), tensor(0.0045...\n",
       "2468    (nv, tensor(5), [tensor(0.0003), tensor(0.0040...\n",
       "2469    (nv, tensor(5), [tensor(0.0003), tensor(0.0040...\n",
       "2470    (nv, tensor(5), [tensor(0.0003), tensor(0.0040...\n",
       "2471    (nv, tensor(5), [tensor(0.0002), tensor(0.0035...\n",
       "2472    (nv, tensor(5), [tensor(0.0002), tensor(0.0035...\n",
       "2473    (nv, tensor(5), [tensor(0.0002), tensor(0.0049...\n",
       "2474    (nv, tensor(5), [tensor(0.0002), tensor(0.0049...\n",
       "2475    (nv, tensor(5), [tensor(0.0002), tensor(0.0047...\n",
       "2476    (nv, tensor(5), [tensor(0.0002), tensor(0.0050...\n",
       "2477    (nv, tensor(5), [tensor(0.0002), tensor(0.0050...\n",
       "2478    (nv, tensor(5), [tensor(0.0002), tensor(0.0050...\n",
       "2479    (nv, tensor(5), [tensor(0.0001), tensor(0.0022...\n",
       "2480    (nv, tensor(5), [tensor(0.0001), tensor(0.0022...\n",
       "2481    (nv, tensor(5), [tensor(0.0002), tensor(0.0049...\n",
       "dtype: object"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "preds"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "p = [x[0].obj for x in preds.values]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[ 0,  0, 20],\n",
       "       [ 0,  0, 20],\n",
       "       [ 0,  0,  0]])"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "confusion_matrix(df.dx, p)"
   ]
  },
  {
   "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": "3.6.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}