SVM Image Recognition.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "competitive-nurse",
   "metadata": {},
   "outputs": [],
   "source": [
    "## Image recognition with Support Vector Machines\n",
    "> We are going to use faces from sklearn"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "abstract-number",
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "from sklearn.datasets import fetch_olivetti_faces"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "subject-jamaica",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'data': array([[0.30991736, 0.3677686 , 0.41735536, ..., 0.15289256, 0.16115703,\n",
       "         0.1570248 ],\n",
       "        [0.45454547, 0.47107437, 0.5123967 , ..., 0.15289256, 0.15289256,\n",
       "         0.15289256],\n",
       "        [0.3181818 , 0.40082645, 0.49173555, ..., 0.14049587, 0.14876033,\n",
       "         0.15289256],\n",
       "        ...,\n",
       "        [0.5       , 0.53305787, 0.607438  , ..., 0.17768595, 0.14876033,\n",
       "         0.19008264],\n",
       "        [0.21487603, 0.21900827, 0.21900827, ..., 0.57438016, 0.59090906,\n",
       "         0.60330576],\n",
       "        [0.5165289 , 0.46280992, 0.28099173, ..., 0.35950413, 0.3553719 ,\n",
       "         0.38429752]], dtype=float32),\n",
       " 'images': array([[[0.30991736, 0.3677686 , 0.41735536, ..., 0.37190083,\n",
       "          0.3305785 , 0.30578512],\n",
       "         [0.3429752 , 0.40495867, 0.43801653, ..., 0.37190083,\n",
       "          0.338843  , 0.3140496 ],\n",
       "         [0.3429752 , 0.41735536, 0.45041323, ..., 0.38016528,\n",
       "          0.338843  , 0.29752067],\n",
       "         ...,\n",
       "         [0.21487603, 0.20661157, 0.2231405 , ..., 0.15289256,\n",
       "          0.16528925, 0.17355372],\n",
       "         [0.20247933, 0.2107438 , 0.2107438 , ..., 0.14876033,\n",
       "          0.16115703, 0.16528925],\n",
       "         [0.20247933, 0.20661157, 0.20247933, ..., 0.15289256,\n",
       "          0.16115703, 0.1570248 ]],\n",
       " \n",
       "        [[0.45454547, 0.47107437, 0.5123967 , ..., 0.19008264,\n",
       "          0.18595041, 0.18595041],\n",
       "         [0.446281  , 0.48347107, 0.5206612 , ..., 0.21487603,\n",
       "          0.2107438 , 0.2107438 ],\n",
       "         [0.49586776, 0.5165289 , 0.53305787, ..., 0.20247933,\n",
       "          0.20661157, 0.20661157],\n",
       "         ...,\n",
       "         [0.77272725, 0.78099173, 0.7933884 , ..., 0.1446281 ,\n",
       "          0.1446281 , 0.1446281 ],\n",
       "         [0.77272725, 0.7768595 , 0.7892562 , ..., 0.13636364,\n",
       "          0.13636364, 0.13636364],\n",
       "         [0.7644628 , 0.7892562 , 0.78099173, ..., 0.15289256,\n",
       "          0.15289256, 0.15289256]],\n",
       " \n",
       "        [[0.3181818 , 0.40082645, 0.49173555, ..., 0.40082645,\n",
       "          0.3553719 , 0.30991736],\n",
       "         [0.30991736, 0.3966942 , 0.47933885, ..., 0.40495867,\n",
       "          0.37603307, 0.30165288],\n",
       "         [0.26859504, 0.34710744, 0.45454547, ..., 0.3966942 ,\n",
       "          0.37190083, 0.30991736],\n",
       "         ...,\n",
       "         [0.1322314 , 0.09917355, 0.08264463, ..., 0.13636364,\n",
       "          0.14876033, 0.15289256],\n",
       "         [0.11570248, 0.09504132, 0.0785124 , ..., 0.1446281 ,\n",
       "          0.1446281 , 0.1570248 ],\n",
       "         [0.11157025, 0.09090909, 0.0785124 , ..., 0.14049587,\n",
       "          0.14876033, 0.15289256]],\n",
       " \n",
       "        ...,\n",
       " \n",
       "        [[0.5       , 0.53305787, 0.607438  , ..., 0.28512397,\n",
       "          0.23966943, 0.21487603],\n",
       "         [0.49173555, 0.5413223 , 0.60330576, ..., 0.29752067,\n",
       "          0.20247933, 0.20661157],\n",
       "         [0.46694216, 0.55785125, 0.6198347 , ..., 0.29752067,\n",
       "          0.17768595, 0.18595041],\n",
       "         ...,\n",
       "         [0.03305785, 0.46280992, 0.5289256 , ..., 0.17355372,\n",
       "          0.17355372, 0.1694215 ],\n",
       "         [0.1570248 , 0.5247934 , 0.53305787, ..., 0.16528925,\n",
       "          0.1570248 , 0.18595041],\n",
       "         [0.45454547, 0.5206612 , 0.53305787, ..., 0.17768595,\n",
       "          0.14876033, 0.19008264]],\n",
       " \n",
       "        [[0.21487603, 0.21900827, 0.21900827, ..., 0.71487606,\n",
       "          0.71487606, 0.6942149 ],\n",
       "         [0.20247933, 0.20661157, 0.20661157, ..., 0.7107438 ,\n",
       "          0.7066116 , 0.6942149 ],\n",
       "         [0.2107438 , 0.20661157, 0.20661157, ..., 0.6859504 ,\n",
       "          0.69008267, 0.6942149 ],\n",
       "         ...,\n",
       "         [0.2644628 , 0.25619835, 0.2603306 , ..., 0.5413223 ,\n",
       "          0.57438016, 0.59090906],\n",
       "         [0.26859504, 0.2644628 , 0.26859504, ..., 0.56198347,\n",
       "          0.58264464, 0.59504133],\n",
       "         [0.27272728, 0.26859504, 0.27272728, ..., 0.57438016,\n",
       "          0.59090906, 0.60330576]],\n",
       " \n",
       "        [[0.5165289 , 0.46280992, 0.28099173, ..., 0.5785124 ,\n",
       "          0.5413223 , 0.60330576],\n",
       "         [0.5165289 , 0.45041323, 0.29338843, ..., 0.58264464,\n",
       "          0.553719  , 0.5785124 ],\n",
       "         [0.5165289 , 0.44214877, 0.29338843, ..., 0.59917355,\n",
       "          0.5785124 , 0.54545456],\n",
       "         ...,\n",
       "         [0.39256197, 0.41322315, 0.38842976, ..., 0.33471075,\n",
       "          0.37190083, 0.3966942 ],\n",
       "         [0.39256197, 0.38429752, 0.40495867, ..., 0.3305785 ,\n",
       "          0.35950413, 0.37603307],\n",
       "         [0.3677686 , 0.40495867, 0.3966942 , ..., 0.35950413,\n",
       "          0.3553719 , 0.38429752]]], dtype=float32),\n",
       " 'target': array([ 0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  1,  1,  1,  1,  1,  1,  1,\n",
       "         1,  1,  1,  2,  2,  2,  2,  2,  2,  2,  2,  2,  2,  3,  3,  3,  3,\n",
       "         3,  3,  3,  3,  3,  3,  4,  4,  4,  4,  4,  4,  4,  4,  4,  4,  5,\n",
       "         5,  5,  5,  5,  5,  5,  5,  5,  5,  6,  6,  6,  6,  6,  6,  6,  6,\n",
       "         6,  6,  7,  7,  7,  7,  7,  7,  7,  7,  7,  7,  8,  8,  8,  8,  8,\n",
       "         8,  8,  8,  8,  8,  9,  9,  9,  9,  9,  9,  9,  9,  9,  9, 10, 10,\n",
       "        10, 10, 10, 10, 10, 10, 10, 10, 11, 11, 11, 11, 11, 11, 11, 11, 11,\n",
       "        11, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 13, 13, 13, 13, 13, 13,\n",
       "        13, 13, 13, 13, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 15, 15, 15,\n",
       "        15, 15, 15, 15, 15, 15, 15, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16,\n",
       "        17, 17, 17, 17, 17, 17, 17, 17, 17, 17, 18, 18, 18, 18, 18, 18, 18,\n",
       "        18, 18, 18, 19, 19, 19, 19, 19, 19, 19, 19, 19, 19, 20, 20, 20, 20,\n",
       "        20, 20, 20, 20, 20, 20, 21, 21, 21, 21, 21, 21, 21, 21, 21, 21, 22,\n",
       "        22, 22, 22, 22, 22, 22, 22, 22, 22, 23, 23, 23, 23, 23, 23, 23, 23,\n",
       "        23, 23, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 25, 25, 25, 25, 25,\n",
       "        25, 25, 25, 25, 25, 26, 26, 26, 26, 26, 26, 26, 26, 26, 26, 27, 27,\n",
       "        27, 27, 27, 27, 27, 27, 27, 27, 28, 28, 28, 28, 28, 28, 28, 28, 28,\n",
       "        28, 29, 29, 29, 29, 29, 29, 29, 29, 29, 29, 30, 30, 30, 30, 30, 30,\n",
       "        30, 30, 30, 30, 31, 31, 31, 31, 31, 31, 31, 31, 31, 31, 32, 32, 32,\n",
       "        32, 32, 32, 32, 32, 32, 32, 33, 33, 33, 33, 33, 33, 33, 33, 33, 33,\n",
       "        34, 34, 34, 34, 34, 34, 34, 34, 34, 34, 35, 35, 35, 35, 35, 35, 35,\n",
       "        35, 35, 35, 36, 36, 36, 36, 36, 36, 36, 36, 36, 36, 37, 37, 37, 37,\n",
       "        37, 37, 37, 37, 37, 37, 38, 38, 38, 38, 38, 38, 38, 38, 38, 38, 39,\n",
       "        39, 39, 39, 39, 39, 39, 39, 39, 39]),\n",
       " 'DESCR': '.. _olivetti_faces_dataset:\\n\\nThe Olivetti faces dataset\\n--------------------------\\n\\n`This dataset contains a set of face images`_ taken between April 1992 and \\nApril 1994 at AT&T Laboratories Cambridge. The\\n:func:`sklearn.datasets.fetch_olivetti_faces` function is the data\\nfetching / caching function that downloads the data\\narchive from AT&T.\\n\\n.. _This dataset contains a set of face images: http://www.cl.cam.ac.uk/research/dtg/attarchive/facedatabase.html\\n\\nAs described on the original website:\\n\\n    There are ten different images of each of 40 distinct subjects. For some\\n    subjects, the images were taken at different times, varying the lighting,\\n    facial expressions (open / closed eyes, smiling / not smiling) and facial\\n    details (glasses / no glasses). All the images were taken against a dark\\n    homogeneous background with the subjects in an upright, frontal position \\n    (with tolerance for some side movement).\\n\\n**Data Set Characteristics:**\\n\\n    =================   =====================\\n    Classes                                40\\n    Samples total                         400\\n    Dimensionality                       4096\\n    Features            real, between 0 and 1\\n    =================   =====================\\n\\nThe image is quantized to 256 grey levels and stored as unsigned 8-bit \\nintegers; the loader will convert these to floating point values on the \\ninterval [0, 1], which are easier to work with for many algorithms.\\n\\nThe \"target\" for this database is an integer from 0 to 39 indicating the\\nidentity of the person pictured; however, with only 10 examples per class, this\\nrelatively small dataset is more interesting from an unsupervised or\\nsemi-supervised perspective.\\n\\nThe original dataset consisted of 92 x 112, while the version available here\\nconsists of 64x64 images.\\n\\nWhen using these images, please give credit to AT&T Laboratories Cambridge.\\n'}"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "faces = fetch_olivetti_faces()\n",
    "faces"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "better-humanitarian",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      ".. _olivetti_faces_dataset:\n",
      "\n",
      "The Olivetti faces dataset\n",
      "--------------------------\n",
      "\n",
      "`This dataset contains a set of face images`_ taken between April 1992 and \n",
      "April 1994 at AT&T Laboratories Cambridge. The\n",
      ":func:`sklearn.datasets.fetch_olivetti_faces` function is the data\n",
      "fetching / caching function that downloads the data\n",
      "archive from AT&T.\n",
      "\n",
      ".. _This dataset contains a set of face images: http://www.cl.cam.ac.uk/research/dtg/attarchive/facedatabase.html\n",
      "\n",
      "As described on the original website:\n",
      "\n",
      "    There are ten different images of each of 40 distinct subjects. For some\n",
      "    subjects, the images were taken at different times, varying the lighting,\n",
      "    facial expressions (open / closed eyes, smiling / not smiling) and facial\n",
      "    details (glasses / no glasses). All the images were taken against a dark\n",
      "    homogeneous background with the subjects in an upright, frontal position \n",
      "    (with tolerance for some side movement).\n",
      "\n",
      "**Data Set Characteristics:**\n",
      "\n",
      "    =================   =====================\n",
      "    Classes                                40\n",
      "    Samples total                         400\n",
      "    Dimensionality                       4096\n",
      "    Features            real, between 0 and 1\n",
      "    =================   =====================\n",
      "\n",
      "The image is quantized to 256 grey levels and stored as unsigned 8-bit \n",
      "integers; the loader will convert these to floating point values on the \n",
      "interval [0, 1], which are easier to work with for many algorithms.\n",
      "\n",
      "The \"target\" for this database is an integer from 0 to 39 indicating the\n",
      "identity of the person pictured; however, with only 10 examples per class, this\n",
      "relatively small dataset is more interesting from an unsupervised or\n",
      "semi-supervised perspective.\n",
      "\n",
      "The original dataset consisted of 92 x 112, while the version available here\n",
      "consists of 64x64 images.\n",
      "\n",
      "When using these images, please give credit to AT&T Laboratories Cambridge.\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(faces.DESCR)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 85,
   "id": "casual-tuner",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(400, 4096)"
      ]
     },
     "execution_count": 85,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "faces.data.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 91,
   "id": "motivated-shark",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0.30991736, 0.3677686 , 0.41735536, ..., 0.15289256, 0.16115703,\n",
       "       0.1570248 ], dtype=float32)"
      ]
     },
     "execution_count": 91,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "faces.data[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "serious-berlin",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(400, 64, 64)"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "faces.images.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "absolute-minneapolis",
   "metadata": {},
   "source": [
    "> We have 400 images of faces"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "temporal-integral",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(64, 64)"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "faces.images[0].shape"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "colonial-platform",
   "metadata": {},
   "source": [
    "> Each image is `64x64` pixcels"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "accompanied-theater",
   "metadata": {},
   "source": [
    "### Visualising the first image"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "mechanical-rochester",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x167adfdbdc0>"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.imshow(faces.images[0], cmap=\"gray\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "increased-worth",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(400,)"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "faces.target.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "quality-hierarchy",
   "metadata": {},
   "outputs": [],
   "source": [
    "> We have 400 different people"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "copyrighted-castle",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "id": "golden-bathroom",
   "metadata": {},
   "source": [
    "#### Training a Support Vector Machine\n",
    "> This is a classification problem\n",
    "\n",
    "> To use SVM in scikit-learn to solve our task, we will import the SVC class from the\n",
    "sklearn.svm module:\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "competitive-creator",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.svm import SVC"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "id": "bizarre-thesaurus",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "SVC(kernel='linear')"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model = SVC(kernel=\"linear\")\n",
    "model"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "informed-suspect",
   "metadata": {},
   "source": [
    "### Splitting our data into trainning and testing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "id": "accessible-island",
   "metadata": {},
   "outputs": [],
   "source": [
    "X = faces.data\n",
    "y = faces.target\n",
    "from sklearn.model_selection import train_test_split\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size = .25, random_state = 29)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "id": "major-judges",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((300, 4096), (100, 4096), (300,), (100,))"
      ]
     },
     "execution_count": 42,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train.shape, X_test.shape, y_train.shape, y_test.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "varying-covering",
   "metadata": {},
   "source": [
    "### Classifying images"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "id": "right-distributor",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "SVC(kernel='linear')"
      ]
     },
     "execution_count": 43,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "id": "filled-correction",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1.0"
      ]
     },
     "execution_count": 44,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.score(X_train, y_train)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "behind-central",
   "metadata": {},
   "source": [
    "> The **model** is `100%` accurate on the train dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "id": "egyptian-tomato",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.94"
      ]
     },
     "execution_count": 45,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.score(X_test, y_test)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "thrown-opposition",
   "metadata": {},
   "source": [
    "> The **model** is `94%` accurate on the test dataset"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "closed-roulette",
   "metadata": {},
   "source": [
    "#### Making some predictions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "id": "linear-reading",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(array([0.4338843 , 0.5206612 , 0.5041322 , ..., 0.5082645 , 0.5041322 ,\n",
       "        0.46694216], dtype=float32),\n",
       " 33)"
      ]
     },
     "execution_count": 46,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train[0], y_train[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "id": "aware-philadelphia",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([33])"
      ]
     },
     "execution_count": 47,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.predict([X_train[0]])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "respective-florida",
   "metadata": {},
   "source": [
    "#### Making predictions on the test data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "id": "fleet-concord",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "34 -----> 34\n",
      "36 -----> 36\n",
      "39 -----> 39\n",
      "26 -----> 26\n",
      "31 -----> 31\n",
      "3 -----> 3\n",
      "18 -----> 18\n",
      "8 -----> 8\n",
      "16 -----> 16\n",
      "29 -----> 29\n",
      "39 -----> 39\n",
      "35 -----> 35\n",
      "29 -----> 4\n",
      "8 -----> 8\n",
      "24 -----> 24\n",
      "6 -----> 6\n",
      "2 -----> 2\n",
      "37 -----> 37\n",
      "32 -----> 32\n",
      "22 -----> 25\n"
     ]
    }
   ],
   "source": [
    "for x, y in list(zip(X_test, y_test))[10:30]:\n",
    "    print(model.predict([x])[0], \"----->\", y)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "proved-jamaica",
   "metadata": {},
   "source": [
    "#### Perfomance"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "id": "fifth-constraint",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.metrics import accuracy_score, f1_score"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "id": "furnished-vocabulary",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.94"
      ]
     },
     "execution_count": 51,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_pred = model.predict(X_test)\n",
    "acc = accuracy_score(y_test, y_pred)\n",
    "acc"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "id": "excess-shame",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1.0"
      ]
     },
     "execution_count": 65,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_pred = model.predict(X_train)\n",
    "acc = accuracy_score(y_train, y_pred)\n",
    "acc"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "subjective-shaft",
   "metadata": {},
   "source": [
    "##### f1_score"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "id": "cardiac-shape",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1.,\n",
       "       1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1.,\n",
       "       1., 1., 1., 1., 1., 1.])"
      ]
     },
     "execution_count": 67,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "\n",
    "f1_score(y_train, y_pred, average=None)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "anonymous-shadow",
   "metadata": {},
   "source": [
    "> The model if predicting Equaly on all faces"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "id": "temporal-treasury",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "dict_keys(['data', 'images', 'target', 'DESCR'])"
      ]
     },
     "execution_count": 62,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "faces.keys()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "welcome-remainder",
   "metadata": {},
   "source": [
    "### Confusion Matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "id": "respiratory-softball",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.metrics import confusion_matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 80,
   "id": "recreational-boards",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[2, 0, 0],\n",
       "       [0, 0, 1],\n",
       "       [1, 0, 2]], dtype=int64)"
      ]
     },
     "execution_count": 80,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_true = [2, 0, 2, 2, 0, 1]\n",
    "y_pred = [0, 0, 2, 2, 0, 2]\n",
    "confusion_matrix(y_true, y_pred)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 81,
   "id": "designed-kingston",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[0, 1, 0],\n",
       "       [0, 2, 1],\n",
       "       [0, 0, 2]], dtype=int64)"
      ]
     },
     "execution_count": 81,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_true = [\"cat\", \"ant\", \"cat\", \"cat\", \"ant\", \"bird\"]\n",
    "y_pred = [\"ant\", \"ant\", \"cat\", \"cat\", \"ant\", \"cat\"]\n",
    "confusion_matrix(y_true, y_pred, labels=list(set(y_true)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "id": "mounted-capital",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[8 0 0 ... 0 0 0]\n",
      " [0 8 0 ... 0 0 0]\n",
      " [0 0 7 ... 0 0 0]\n",
      " ...\n",
      " [0 0 0 ... 8 0 0]\n",
      " [0 0 0 ... 0 6 0]\n",
      " [0 0 0 ... 0 0 7]]\n"
     ]
    }
   ],
   "source": [
    "y_pred = model.predict(X_train)\n",
    "print(confusion_matrix(y_train, y_pred))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "accredited-walnut",
   "metadata": {},
   "source": [
    "> Done"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "informed-prison",
   "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": "3.8.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}