svm-example.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "from sklearn import datasets\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "from sklearn.svm import LinearSVC\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "def plot_decision_boundary(clf, axes):\n",
    "    x1 = np.linspace(axes[0], axes[1], 100)\n",
    "    x2 = np.linspace(axes[2], axes[3], 100)\n",
    "    \n",
    "    x1_mesh, x2_mesh = np.meshgrid(x1, x2)\n",
    "    x1_x2 = np.c_[x1_mesh.ravel(), x2_mesh.ravel()]\n",
    "    \n",
    "    grid_prediction = clf.predict(x1_x2).reshape(x1_mesh.shape)\n",
    "    plt.contourf(x1_mesh, x2_mesh, grid_prediction, alpha=0.3)\n",
    "    \n",
    "def plot_dataset(X, y, axes):\n",
    "    plt.plot(X[:, 0][y==0], X[:, 1][y==0], \"bs\")\n",
    "    plt.plot(X[:, 0][y==1], X[:, 1][y==1], \"g^\")\n",
    "    plt.axis(axes)\n",
    "    plt.grid(True, which='both')\n",
    "    \n",
    "def get_linearly_separable_iris_data(scale=True):\n",
    "    iris = datasets.load_iris()\n",
    "    X = iris['data'][:, (2, 3)]\n",
    "    y = (iris['target'] == 2).astype(np.float64)\n",
    "    idx = []\n",
    "    \n",
    "    # making sure the data is linearly separable so we can fit a hard-margin svm\n",
    "    for i, x in enumerate(X):\n",
    "        if y[i] != 0 or x[0] < 4.2:\n",
    "            idx.append(i)\n",
    "    \n",
    "    X, y = X[idx], y[idx]\n",
    "    \n",
    "    if scale:\n",
    "        scaler = StandardScaler()\n",
    "        X = scaler.fit_transform(X)\n",
    "            \n",
    "    return X, y"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[2.18464802 2.27591335]\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAmMAAAHWCAYAAAA7J44mAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjMsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+AADFEAAAgAElEQVR4nO3df3Bc533f+88XPyhIEAmQAgmKgCBEKCYiDF6xji4lOx6NU8U/QrG100kqN3MbK22Gk0w8/XF1Z+pbd5KZe29vKnVuZ5w6lqNp3NoJIzszjRuVUiI5aFyJE/6SbFJcAbL5owSwJAgQS5EgCK0Wi33uH4uzXAIHi8X+OrvnvF8zHC3Os3vOcx4cUV89z7PfrznnBAAAgGA0Bd0BAACAKCMYAwAACBDBGAAAQIAIxgAAAAJEMAYAABAggjEAAIAAlR2MmdkDZvbXZjZmZu+a2T/zeY+Z2e+Z2Tkze8fMPlrudQEAAMKgpQLnSEt61jn3QzPbLOltM/u+c2407z2/IGlw+c9jkl5Y/icAAECklT0z5pybcs79cPn1TUljknpWvO1zkr7tso5J6jSz+8u9NgAAQKOr6J4xM+uX9LclHV/R1CNpMu/nuFYHbAAAAJFTiWVKSZKZ3Svpv0j65865uZXNPh/xrcNkZgclHZSktra2n+nt7a1UF0PBOSczv+GsnaWMy/7yTDJzajInk6nJgvs+SCbj1NQU7LjUI8bFH+OyGmPij3Hxx7j4+8lPzs0657Zv9HMVCcbMrFXZQOyQc+7PfN4Sl/RA3s+9ki77ncs596KkFyVpcHDQ/dHX/U4XXYlkXPe1BR+gjp6f1oWbc5p44Ja6dl3Xkw/eUF/7feq7528F0p+3jiX16ONtgVy7njEu/hiX1RgTf4yLP8bFX2f7p8dL+Vwlvk1pkv5Q0phz7t+v8baXJf3q8rcqH5d0wzk3Ve61EZyhgW49tHmL+ibb1TTRr5GzQzo5M64jV1euUAMAgEIqMTP2s5L+kaQzZnZq+di/ktQnSc65b0h6VdJ+SeckLUj6tQpcFwEbGuiWzksXpuYUn15QTPs0s31O0vFAZ8kAAGgkZQdjzrkj8t8Tlv8eJ+m3yr0W6s/QQLeG1K3R89N64/QVzfY0S7t7lV2ZFgEZAADrqNgGfkRbbpZsck6z6VaN9N3UcNe4tIOADACAQgjGUDH5y5ZJ9SuW6pF0QhPtCZYtAQBYA8EYKspbtjx86uyKfWQsWwIA4IdC4aiKA3sH9UT7Tt3zI2l0rFkj4x2auJXQxMK5oLsGAEBdYWYMVeMtW2pSSqY7NZJa1PCu7LLlJ7ZTmhQAAIlgDFV25z6yVtJfAACwAsEYqu7O9Bc3SH8BAEAegjHUDOkvAABYjWAMNUX6CwAA7kQwhpoj/QUAALeR2gKBIf0FAADMjCFgq9Jf6KKGu8ZJfwEAiAyCMQTObx8Z6S8AAFFBMIa6QPoLAEBUEYyhruSnvxjVLc3s6iD9BQAg1AjGUHdW7iMj/QUAIMwIxlCXSH8BAIgKUlugrnnpL66dXiD9BQAglJgZQ93L30dG+gsAQNgQjKEhrJf+QuoNuosAAJSEYAwNo1D6i6ZMl6S2oLsIAMCGEYyh4XizZKnWNv3k6ozOqFe7M+mguwUAQEnYwI+G1teyQ3erP+huAABQMoIxAACAABGMAQAABIhgDA0rfmHmjp/JPQYAaERs4EdD8jbxv/H2RWV6nLo77tHErYQksvMDABoLwRgaVn4y2PSmbHZ+iooDABoNwRgamheQNS3Oq2liORns4pz2dB6nqDgAoCEQjKHhDQ10K5FcVNt7Hyg+vSA9skVnlE0GKzFLBgCob2zgR2h4RcXv+ZEoKg4AaBjMjCFUvGVLTSpbVDy1qOFdJygqDgCoWwRjCJ07i4q3KqZ9dxQVZ9kSAFBPCMYQSoWKikvsIwMA1A+CMYRafvqL2XSrRvpukv4CAFBXCMYQencuW2bTX0jZfWQsWwIAgkYwhkjwli0Pnzqr+PRC3j4yli0BAMEitQUihfQXAIB6w8wYImdV+gtd1HDXOOkvAACBYGYMkTQ00K2HNm9R29RitozS5X06c71XR64eZ5YMQMVcmU9o/0vPanr+Ws3OW61ronoIxhBZQwPdOrB3UEPzd+va6QWNjjXrzPVeli0BVMzzRw/pWDym548eqtl5q3VNVA/BGCJvaKBbT7TvVN9ke24f2cmZcQIyAGW5Mp/Qn8ReV0ZOh2KvVWymqtB5q3VNVBfBGKDby5Z9k+25ZcuTM+MsWwIo2fNHDynjMpKkjMtUbKaq0HmrdU1UF8EYsMxbtmybWtS10wu5fWQsWwLYKG+GKrWUliSlltIVmakqdN5qXRPVRzAGrED6CwDlyp+h8lRipqrQeat1TVQfqS0AH6S/AFCOk5fGcjNUntRSWicujVb1vNW4JqqPYAxYg18ZpWzW/uOUUQJQ0JvPvNBQ50WwCMaAArwySqPnp/XG6Rua7WmWdveKMkoAgEohGAOKkJslm5zTqG5pZleHhrvGpR0EZACA8hCMAUVauY8sluqRdEIT7QmWLQEAJSMYAzbAW7Y8fOqs4tMLimnf8j4yli0BAKUhtQVQAi/9hVdGifQXAIBSMTMGlCh/HxnpLwAAparIzJiZfdPMZswstkb7J83shpmdWv7z25W4LhA0r4xS29RirozSmeu9lFECABStUsuU/1nSZ9d5z5vOub3Lf/6vCl0XCNzKMkqjY82UUQKq5Mp8QvtferauSvy8M31OfV/9vGIzF4LuChpURYIx59wbkurn3wwgAN4+sr7J9tw+spMz4wRkQAU9f/SQjsVjdVXi5+Arz2kutaBfP/y7QXcFDaqWG/g/ZmanzewvzOwjNbwuUDPesmXfZHtu2fLkzDjLlkAFeIWwM3J1UwD7nelzei8xLkl6LzHO7BhKYs65ypzIrF/SYefcsE/bFkkZ59y8me2X9FXn3OAa5zko6aAkbd++/Wf+0x9+qyL9C4t0ZlEtTa1Bd6Pu1Nu4JJNpfZhZUkoZtdzTpJaWJd3dnNKmphZtamqrWT8WbmV0Tztfml6JcVmtEcbkq+de0F9c+b7SLq0Wa9H+nZ/SP/1bv1nVa643Lv/k7d/S+MJk7uf+e/r0H3/ma1XtUz1ohOclCD//5N972zn36EY/V5NvUzrn5vJev2pmXzezLufcrM97X5T0oiQNDg66+9p6a9HFhpFIxsWYrFZ347Icb42en9Zfxa8o0+PUteu6nnzwhnbUMEHsW8eSevTx2gV/jYJxWa3ex+TKfELfPzqitMsWwk67tF6/OqJ/9/kvqvvebVW7bqFxeWf6nMbfnLzj2MWFCbU9dFnDOx6qWp/qQb0/L42mJmGtme00M1t+vW/5uolaXBsI0tBAd24fWdNEf24f2ZGrx4PuGtBQnj96SBmXueNYxmUC3Tt28JXnfI+zdwwbVZGZMTN7SdInJXWZWVzS70hqlSTn3Dck/ZKk3zSztKQPJH3BVWp9FKhzuXxkU3NKql+xVM9y1v7jlFECinTy0phSS+k7jqWW0jpxaTSgHkkXr09t6DiwlooEY865f7hO+9ckhX8RHVjDyjJKsz3N0u5eUUYJKM6bz7wQdBdWufK/Hw66CwgJMvADNXRg76BGz0/rwuScRnVLM7s6NNw1Lu0gIAOAqCIYA2rMW7bUpJRMdyqW6pF0QhPtCZYtASCCCMaAAOTvI4tPLyimfcv7yFi2BICoIRgDAuLtIxs9P603Tl/RbE+zZnZ16MkHs180JiADgGggGAMClpslm5xTMt2pEV3UcNe4JtoT+sT2x4LuHgCgygjGgDpA+gsAiC5qGQB1YmigWwf2DqptalHXTi9odKxZZ673auJWgrqWwLIr8wntf+nZNetSFmovta2a/a3GZ6t1L/UmTPdJMAbUmQN7B3NZ+0fHmnNZ+wnIgGwm/mPx2JqZ9wu1l9pWzf5W47PVupd6E6b7JBgD6tDQQLce2rwlV0YpdnkfZZQQeVfmE/qT2OvKyOlQ7LVVMyKF2kttq2Z/q/HZat1LvQnbfRKMAXXKC8i8ZcvY5X06c71XR64eZ5YMkZRfn9KvLmWh9lLbqtnfany2WvdSb8J2nwRjQB3z9pE90b4zt49sZLyDfWSIHG8mxKtPmVpK3zEjUqi91LZq9rcan63WvdSbMN4nwRjQAIYGunP7yJom+tlHhsjJnwnx5M+IFGovta2a/a3GZ6t1L/UmjPdJagugQfimv1ic055O0l8g/E5eGsvNhHhSS2mduDRaVHupbdXqbzU+W841G0kY75NgDGggXtb+w6fOKj69oNmeZml3ryijhLB785kXymqvtXL6U+pn620MqiWM90kwBjSgA3sHNXp+Whcm5zSqW5RRAoAGRjAGNChv2VKTypZRSi1qeNcJyigBQINhAz/QwPLTXzRdaVXs8j7Naw8b+wGggTAzBjQ4b4YsdXebfqKZoLsDANggZsYAAAACRDAGAAAQIIIxIGT+5/X3g+4CAGADCMaAkIhfmFHTlVbNXN2SKyqeyiSD7hYQelfmE9r/0rMbLsdT6ufChnEgGANCwathubKoeCqT5puVQJU9f/SQjsVjGy7HU+rnwoZxIBgDQmVlUfGbqWaKigNV5BWtzshtqFh1qZ8LG8Yhi2AMCJn8ouJK3ZUrKn7k6vGguwaETn7R6o0Uqy71c2HDOGQRjAEh5CWDbVp0aprozy1bHrl6nFkyoEK8WR2vaHVqKV3U7E6pnwsbxuE2gjEgpIYGutVxz125fWSjY806c72XZUugQvJndTzFzO6U+rmwYRxuIwM/EHJ+RcWHu8alHRQVB8px8tJYblbHk1pK68Sl0ap8LmwYh9sIxoAIWFlUPJbqkZQtKt7Xfh9BGVCCN595oaafCxvG4TaCMSAivIDswtSc4tMLimmfZrbPSYpLYpYMAIJCMAZEyNBAt4bUrdHz03rj9BXN9jRrZleHnnwwIYmADACCQDAGRFBulmxyTsl0p0Z0UcNd45poT+gT2x8LunsAECkEY0BE5S9bJtWvWKpnednyOPvIAKCGCMaACPOWLQ+fOqv49IJme5ql3b1iHxkA1A7BGADSXwBAgAjGAEgqnP6CfWQAUD1k4AeQ45VR8rL2U0YJUXFlPqH9Lz3rW4qnUFu5ny21T40kLPdRTQRjAO4wNNCtA3sH9UT7zlwZpZHxDsooIdSeP3pIx+Ix31I8hdrK/WypfWokYbmPaiIYA+BraKBbT7TvVN9ku5om+jUy3qGTM+MEZAgdr2B1Rm5VoepE6tqabet9tlBbOX1qJGG5j2ojGAOwpvxly6aJfsUu79Mrl+5m2RKhkl+wemWh6j+e+O6abet9tlBbOX1qJGG5j2ojGANQkLds6e0jGx1r1pnrvSxbIhS8mRuvYHVqKZ2bwbkyn9Br0yO+bcV8dq22cvrUSMJyH7VAMAagKN4+sr7JdvaRITTyZ2483gxOobZyP1tqnxpJWO6jFkhtAaBoK9NfjKQWNbyL9BdoXCcvjeVmbjyppbROXBqVJKXd2m3rfbZQWzl9ahRhuY9aIBgDsCH5ZZTi0wuKaR9llNCw3nzmhYLtbx1L6tHH20r6bKmqdd5aC8t91ALBGIAN88oojZ6f1hunr2i2p1kzuzr05IMJSWTtB4CNIBgDULLcLNnkXHbZUhcpowQAG0QwBqAs+cuWSfUrlurRzOKc9nSybAkAxSAYA1A2b9ny8Kmzik8vaLanWdrdKykuiVkyACiEYAxAxRzYO6jR89O6MDmnUd1iHxkAFIFgDEBFkf4CADaGYAxAxZH+AgCKRwZ+AFXhlVF6on0nZZSwypX5hPa/9KxvaZxS24ppR+mqNbb8zgjGAFTZ0EB3rozS7Jn7NDLeoZMz4wRkEff80UM6Fo/5lsYpta2YdpSuWmPL76xCwZiZfdPMZswstka7mdnvmdk5M3vHzD5aiesCaAxDA916aPMWtU0tqmmiX7HL+/TKpbt15OpxgrII8gpIZ+RWFY4uta2YdpRu5dheS71flfNG9XdWqZmx/yzpswXaf0HS4PKfg5KokQBEjLds2Ta1qGunFzRzdQvLlhGVX0B6ZeHoUtuKaUfpVo7tH098pyrnjervrCLBmHPuDUmFwtnPSfq2yzomqdPM7q/EtQE0Fm8f2T0/kkbHmjUy3kFAFiHeTIhXQDq1lM7NiJTatt55UR6/sf3L6ZGyx5bf2W212jPWI2ky7+f48jEAEeQtW/ZNtqtpol8jZ4d0cmZcR64eD7prqLL8mRCPNyNSatt650V5qjW2/M5uq1VqC/M55nzfaHZQ2aVMbd++XYlkvJr9ajjpzCJj4oNx8VfP49LdI3UkW/VhZkmZuWbpg8d1s2VJP2ie1aamFm1qaqvatRduZfTWsWTVzt+IajUm/+Psu7mZEE9qKa0fnI3lXm+07a3NyYLnfWtz6ffFs+L/O0u78se2Wr+zRmTO+cZEGz+RWb+kw865YZ+2P5D0A+fcS8s//1jSJ51zU4XOOTg46P7o639Wkf6FRSIZ131tvUF3o+4wLv4aZVxGz0/rjVtXlOlxGtq9pD2d8armI3vrWFKPPl69YK8RMSb+GBd/jIu/zvZPv+2ce3Sjn6vVzNjLkr5kZt+R9JikG+sFYgCiI5ckdnJOs+lWjfTd1HDXuLSDMkoAwq8iwZiZvSTpk5K6zCwu6XcktUqSc+4bkl6VtF/SOUkLkn6tEtcFEB75WfuT6lcs1SMpW0aJrP0AwqwiwZhz7h+u0+4k/VYlrgUgvIYGujWkbh0+dXZFGaXsvjcCMgBhRAZ+AHWH9BcAooRC4QDqkrdsqUkpme7UiC5quGtcE+0JfWL7Y0F3DwAqhmAMQN3y20eWXbY8zj4yAKFBMAagrnn7yEbPT+uN0zc029Ms7e4V+8gAhAV7xgA0hKGBbj3RvlN9k+25fWQnZ8bZRxZCV+YT2v/Ss75lcd6ZPqe+r35esZkLFT1vmETlPsOEYAxAw1hZRil2eV+ujBJBWXg8f/SQjsVjvmVxDr7ynOZSC/r1w79b0fOGSVTuM0wIxgA0lKGBbh3YO6i2qUVdO72g2OV9OnO9l29bhoRXPDojt6po9DvT5/ReYlyS9F5ifEOzY4XOGyZRuc+wIRgD0JC89BfXTi+Q/iJE8otHrywaffCV5+5470ZmxwqdN0yicp9hQzAGoGHl7yNrmujP7SM7cvV40F1DCbxZHa94dGopnZvdyZ8V8xQ7O1bovGESlfsMI4IxAA3N20fWNrWY20d25nov+8gaUP6sjseb3Vk5K+YpZnas0HnDJCr3GUaktgDQ8Eh/EQ4nL43lZnU8qaW0Tlwa1cXrU76fWet4secNk6jcZxgRjAEIDS9JbKq1TT+5OqMzIiBrJG8+80JDnbfeROU+w4hgDEAo9bXskGmb5rVV0uWguwMAa2LPGAAAQIAIxgAAAAJEMAYgdOIXZoLuAgAUjT1jAELF28T/xtsXlelxGtq9pHuVkMQmfgD1iWAMQOh4AdmFyTnNpls10ndTw13j0g4CMgD1h2AMQCjlArKpOSXVr1iqRzOLc9rTeVxNmZ+W1BZ0FwFAEnvGAITYyqLiM1e36Mz1XqUyabLz18A70+fU99XP+5YsKtRWyJX5hPa/9GxNS/ys19dS+1TOvVRrHIIYXxCMAYgAr6j4PT+SRseadTPVTFHxGjj4ynOaSy34liwq1FbI80cP6Vg8VtMSP+v1tdQ+lXMv1RqHIMYXBGMAIsKrYdk32S6l7tLI2SGKildRfmHvlQW9C7UV4hXCzsjVrAD2ufkLBftaap/KuZdqjUMQ44ssgjEAkeEFZE2LTk1XWikqXkUrC3vnzyoVaiskvxB2rQpg/+6P/787fl7Z11L7VM69VGscghhfZBGMAYiUoYFuddxzl4bm79a10wsaHWvWmeu9LFtWUP7Ml8ebVSrUVog3a+MVwk4tpas+e/PO9DmNL0yu2ddS+1TOvVRrHIIYX9xGMAYgkoYGuvVE+071TbZr9sx9Ghnv0MmZcQKyClg58+X59cO/W7CtkPxZG0+1Z2/W62upfSrnXqo1DkGML24jtQWAyPJLfyGd0ER7Qn3t95GTrEQXr09t6Ph6bZJ08tJYbtbGk1pK68Sl0Y13sEjr3UepfSrnXqo1DkGML24z51zQfVjT4OCg+6Ov/1nQ3agriWRc97X1Bt2NusO4+GNc/PmNy+FTZxVvWtC2R+7Rju1z2tMZj1RA9taxpB59nNxrKzEu/hgXf53tn37bOffoRj/HMiUAaHX6i5HxDvaRAagJlikBYJm3bKlJKZnu1IguarhrXBPtCX1i+2NBdw9ASBGMAUAe3zJK2+ckHY/UsiWA2iEYA4AVhga6NaRujZ6f1hunb2i2p1na3SspLoli4wAqi2AMANaQmyWbnNOobmlmV4eGu8alHQRkACqHYAwACli5j4z0FwAqjW9TAsA6hga6dWDvoNqmFnXt9EKujFJYvm15ZT6hf3H6/yTbOhAQgjEAKFJY0188f/SQYnOjZFsHAkIwBgAb4BUb75tsV9NEf66M0pGrx4PuWkm8moROjlqEQEAIxgBgg7yArG1qUU0T/bllyyNXjzfcLFl+TUJqEQLBIBgDgBJ4+8iG5u/WtdMLGh1rbrh9ZN6smFeTMLWUZnYMCADBGACUYWigW0+071TfZHtuH9nJmfGGCMjyZ8U8zI4BtUdqCwAoU6Omvzh5aSw3K+ZJLaV14tJoQD0CoolgDAAqwMvaf/jUWcWnFxTTvuUySvWbtf/NZ17IvX7rWFKPPt4WYG+A6GKZEgAqyEt/4e0jC0v6CwDVw8wYAFRYfhmlZLpTI7qo4a5xTbQn9IntjwXdPQB1hmAMAKogF5BNzSmpfsVSPcvLlsfreh8ZgNojGAOAKlm5j2y2p1na3at63kcGoPYIxgCgyg7sHdTo+WldmJzTqG5pZleHhrvGpR0EZAAIxgCgJho1/QWA6iMYA4Aayd9H1ijpLwBUH6ktAKCGvDJKpL/AlfmE9r/0LOWnQDAGAEHIL6PUNNGfK6N05OrxoLuGGnn+6CEdi8coPwWCMQAIytBAtx7avEVtU4tqmuhX7PI+nbneqyNXjzNLFnJekfaMHMXZQTAGAEHyli3bphZzy5ZnrveybBly+UXaKc4OgjEAqAPePrK+yfbcPrKTM+MEZCHkzYp5RdpTS2lmxyKOYAwA6oS3bOntI4td3pfbR0ZQFh75s2IeZseirSLBmJl91sx+bGbnzOzLPu2fNLMbZnZq+c9vV+K6ABA2+fvIrp1eyO0jY9kyPE5eGsvNinlSS2mduDQaUI8QtLLzjJlZs6Tfl/QpZZPlnDSzl51zK5+qN51zB8q9HgCEnVdGafT8tN44fUWzPc2a2dWhJx9MSCIfWaN785kXgu4C6kwlkr7uk3TOOXdBkszsO5I+J4kQHwDKkEsSOzmnZLpTI7qo4a5xTbQn9IntjwXdPQAVUolgrEfSZN7PcUl+f0t8zMxOS7os6f9wzr3rdzIzOyjpoCRt375diWS8Al0Mj3RmkTHxwbj4Y1z8NdK4dPdIHclWfZhZUma2T7rxgG62LOkHzbPa1NSiTU1tFbnOwq2M3jqWrMi5woRx8ce4VFYlgjHzOeZW/PxDSQ865+bNbL+k/ypp0O9kzrkXJb0oSYODg+6+tt4KdDE8Esm4GJPVGBd/jIu/hhuX5Xjr8KmzijctKNPjNLR7SXs6z2tHhepavnUsqUcfr0xgFyaMiz/GpbIqsYE/LumBvJ97lZ39ynHOzTnn5pdfvyqp1cy6KnBtAIgM0l8A4VSJmbGTkgbN7KckXZL0BUm/kv8GM9spado558xsn7JBYKIC1waASPH2kWlSSqY7FUv1SDrBPjKggZUdjDnn0mb2JUmvSWqW9E3n3Ltm9hvL7d+Q9EuSftPM0pI+kPQF59zKpUwAQBFyG/un5hSfXlBM+zSzfU7ScfVVaNkSQO1UYmbMW3p8dcWxb+S9/pqkr1XiWgAA0l8AYUIGfgBoYEMD3bl9ZE0T/aHYR3ZlPqH9Lz3rWx6oUBvQqAjGAKDB5Wft98oovXLp7oYto/T80UM6Fo/5lgcq1AY0KoIxAAiBoYFuHdg7mCujNDrW3JBllLwi2hm5VcWzC7UBjYxgDABCxC/9RSMFZPlFtFcWzy7UBjQygjEACBlv2TK3j+zskE7OjOvI1eNBd60gb+bLK6KdWkrnZsAKtQGNjmAMAELojn1kV1oVu7wvu2xZxzNk+TNfHm8GrFAb0OgqktoCAFB/vHxkqbvblGyRkpoLuksFnbw0lpv58qSW0jpxaTT3eq02oJERjAEA6sKbz7wQdBeAQLBMCQAAECCCMQAAgACxTAkAIRe/MKPkB63KpLfoZOt7mmhPUMMSqCMEYwAQYl4Ny8Onzq4oKh6XRA1LoB6wTAkAEeAlg/Wy8zdaMlggzJgZA4CI8FJdXJicUzLdqRFd1HDXuCbaE2rTI0F3D4gsZsYAIEL8ioqfud6r+fQ8s2RAQJgZA4AKeerph3Xt/dZVx7dtXdQr330vgB75W7mPbLanWf0dmzRxKyGJfWQbdWU+oX/83/5f/ae/+xV137st6O6gATEzBgAV4heIFToetPyi4smkaWS8Qydnxpkh26Dnjx7SsXiM0kwoGcEYAESYt2y5KdWUW7b0iooTlK3PK2CekaNwOUpGMAYAETc00K27mprVNrWoa6cXbhcV59uW68ovYE7hcpSKYAwAoLa2FtJfbJA3K+YVME8tpZkdQ0kIxgAAOUMD3bl9ZE0T/bl9ZEeuHg+6a3Unf1bMw+wYSsG3KQGgQrZtXVzz25SNJJePbGpOSfUrlupZztp/nDJKeU5eGsvNinlSS2mduDQaUI/QqAjGAKBC6il9Rbn80l9od68oo3Tbm8+8EHQXEBIEYwCANR3YO6jR89O6MDmnUd3SzK4ODXeNSzsIyIBKIRgDABTkLVtqUkqmOxVL9Ug6oYn2hD6x/bGguwc0PIIxAMC68veRxacXFNM+9pEBFUIwBgAoirePbPT8tN44fUWzPc2a2dWhJx+kjBJQDoIxAMCG5GbJJueUTHdqRBfZRwaUgWAMALBhvukvFue0p5NlS2CjCNYEadUAABqFSURBVMYAACUh/QVQGQRjACLnf/vVv6Pr1+9adXzb1sWq5Qr72KeHJZlPi9PR12P6+GeG5dzqdjOnv3ktVpU+VQrpL4DyEIwBiBy/QEySb/b8yvELxG4f9wvECh2vN6S/AEpHMAYAqAjSXwClIRgDAFQM6S+AjSMYAwBUHOkvgOIRjAEAqoL0F0BxCMYARE5n54drfpuyepzW+jallP3W5FrfpmxkpL8A1kcwBiBy/vjb/133tfXW9JpHXy+cnqLe01eUyy/9BfvIgCyCMQBATaxMfzGSWtTwLtJfAARjAICaIf0FsBrBGACgpvzSX7CPDFFGMAYACER++ovZdKtG+m6S/gKRRDAGAAgM6S8AgjEAQMBWpr/QI1t0RixbIjoIxgCgQp56+mHfYuNe/rK12l757ntV71sjyKW/+NGcRh+gjBKig2AMACrEL9gqdHy9tigi/QWiiGAMAFBX7txH1kr6C4QewRgAoO7cmf7iBukvEGoEYwCAukX6C0QBwRgAoK75pb+QsvvIWLZEGBCMAUCFbNu6WNK3KbG+lekvbu8jY9kSjY9gDAAqhBQV1Uf6C4QRwRgAoKGsSn+hixruGif9BRpWUyVOYmafNbMfm9k5M/uyT7uZ2e8tt79jZh+txHUBANE0NNCthzZvUdvUopom+hW7vE9nrvfqyNXjmlg4F3T3gA0pe2bMzJol/b6kTym7eH/SzF52zo3mve0XJA0u/3lM0gvL/wQAoCSkv0BYVGKZcp+kc865C5JkZt+R9DlJ+cHY5yR92znnJB0zs04zu985N1WB6wMAIiw//cWobmlmVwfpL9BQKhGM9UiazPs5rtWzXn7v6ZG0Khgzs4OSDkrS9u3blUjGK9DF8EhnFhkTH4yLP8bFH+OyWqOPSXeP1JFs1VBmszKzW6QbD+gnE7d0oWlWm5patKmpraTzLtzK6K1jyQr3tvExLpVViWDMfI65Et6TPejci5JelKTBwUF3X1tveb0LmUQyLsZkNcbFH+Pij3FZLRRjshxvHT51VvGmBW175B7t2D6nPZ3ntaPEfGRvHUvq0cdLC+TCjHGprEps4I9LeiDv515Jl0t4DwAAZTuwd1BPtO/UtdMLGh1r1sh4hyZuJdjYj7pViZmxk5IGzeynJF2S9AVJv7LiPS9L+tLyfrLHJN1gvxgQLk89/fCaSU3DlH+r0H1Kayd2DdMYNIL8fWSkv0C9KzsYc86lzexLkl6T1Czpm865d83sN5bbvyHpVUn7JZ2TtCDp18q9LoD64heEFDreqEq5z7CNQaPwK6OUzdp/nDJKqCsVSfrqnHtV2YAr/9g38l47Sb9ViWsBAFAs0l+gEZCBHwAQeivTXxCQoZ4QjAEAIsELyJLpVt2tzZrXVvFdMtSDipRDAgAAQGkIxgBUhPdtwmKPN6pC9xmVMQBQWSxTAqiIqKRuiMp9AqgdgjEAQKTMTt/QTMt1De1e0r1KSGITP4JFMAagrrUuzeiuzBU1uQ+Vsbv0YdNOLTbvCLpbaFD536qcTbdqpO8mRcUROIIxAPXHpXVv+oy6kq/o3sV35axZ2XK2JnNLmm/9iGbbntJ8yx7J+GsMG+ObDHZxTns6SQaLYPC3GIC60pK5pv7553TX0oSkFi02bZPMbr/BObWnf6z2+Xf1YXOfLt77ZaWbtgbWXzQmLxns4VNnFZ9ekB7ZojMi9xiCwbcpAdSNlsw1Ddz817prKa60bVO6qePOQEySzJRu6lDatumupbgGbv5rtWSuBdNhNDyvqPg9PxJFxREYZsYA1AeXVv/8c2rJ3ChupstMaduqlsz76p9/Tuc2/xuWLFESb9lSk8oWFU8tanjXCU20J9SmR4LuHiKAmTEAdeHe9BndtTShtHVu6HNp69RdSxO6Nx2rUs8QBUMD3Xpo8xa1TS2q6UqrYpf36cz1Xs2n55klQ9Xxv5EA6kJX8hVJLXcsS973D/5azq3+f0azjBJ/+nPeD5JrVlfysOZb9+be8/HPDMs58/ms03/783jBvjz19MO69n7rquPbti6WnGdsvXNW45rYGL+i4v0dmzRxi/QXqC6CMQCBa12a0b2L72Y36+fxC8T8jqetQ/cuvqvWpZlc2gu/QKzQ8Xx+QVGh48VY75zVuCZKk5/+Ir0pu4+M9BeoJoIxAIG7K3Mlm75i5Wb9YpnJWbM2ZabJQYaK8AKypsV5NU1k019I2X1kpL9ApRGMAQhck/tQ2TxilTgPUBlDA91KJBfV9t4Hik8vKKZ9mtk+J9JfoNLYwA8gcBm7S1KJs2KrzgNUFukvUG3MjAEI3IdNO2VuSXKutKVK52RuSamm7sp3DpBP+gtd1HDXuCbaE/rE9seC7h4aHDNjAAK32LxD860fUYubu+O4Wcb3/SuPt7gbmm/9yB37xcz8lz3XOp5v29bFDR0vxnrnrMY1UVl3pL+Y6M+lvzhy9TizZCgLM2MA6sJs21Nqn3/3jtmxXPqKQpyTtKTZtgN3HP6b19bOO5ZIFj5lNVJJrHdO0lc0Br/0F9pNGSWUh2AMQF2Yb9mjD5v7lkshFV9rssXd0IfNfZpvGa5i74A75ae/mE23aqTvJukvUDKWKQHUB2vRxXv/pdJNnWrJvL8841WAc2rJvK90U4cu3vtlSiGh5vyWLU/OjLNsiQ0jGANQN9JN23R+8/+tD5t71eKuqSVzfXVQ5pxaMtfV4q7pw+Zend/8/xRXyxKogqGBbh3YO6i2qUVdO72Q20fGty2xEfyvJIC6km7apnOb/43uTcfUlTysexffzSaEXWZuSfOtH9Fs24Hs0iQzYqgDB/YOavT8tC78aE6jDzRrZleHnnyQMkooDn+LAag/1qL51r2ab92r1qUZbcpMq8l9qIzdpVRTN1n2UZdIf4FSEYwBqGuLzTsIvtAwchv7p+aUVLaMUjZr/3HKKGFNBGMAAFQQ6S+wUQRjAABUQX76i1Hd0syuDtJfwBfBGAAAVbJyH1ks1SPphCbaEyxbIodgDACAKvKWLQ+fOqv49IJi2re8j4xlS2SRZwwAgBo4sHdQT7Tv1LXTCxoda9bIeAf5yCCJmTEAAGomfx8Z6S/gIRgDAKCGSH+BlQjGAACosZX7yEh/EW0EYwAABCRXRon0F5FGMAYAQIBIfwGCMQAAApa/j4z0F9FDMAYAQB24s4zSFc32NGtmV4eefDAhiYAszAjGAACoI6S/iB6CMQAA6gzpL6KFYAwAgDpE+ovoIBgDAKCOkf4i/AjGAACoc6S/CDeCMQAAGgDpL8KLYAwAgAZB+otwIhgDAKDBkP4iXAjGAABoQKS/CA+CMQAAGhTpL8KBYAwAgAZH+ovGRjAGAEAIFEp/wT6y+kYwBgBASKyd/oJ9ZPWMYAwAgBAh/UXjIRgDACCE1kp/wT6y+lNWMGZm2yR9V1K/pIuS/oFz7n2f912UdFPSkqS0c+7Rcq4LAADW55v+YnFOezpZtqwnTWV+/suSRpxzg5JGln9ey8855/YSiAEAUDtDA906sHdQbVOLunZ6QaNjzTpzvVcTtxKaWDgXdPeg8oOxz0n61vLrb0n6fJnnAwAAVXBg76CeaN+pvsl2jY41a2S8g4CsTpS7Z6zbOTclSc65KTPbscb7nKTXzcxJ+gPn3ItlXhcAAGzQyvQXI6lFDe8i/UXQ1g3GzOyvJO30afrKBq7zs865y8vB2vfN7D3n3BtrXO+gpIOStH37diWS8Q1cJvzSmUXGxAfj4o9x8ce4rMaY+AvjuHT3SB3JVn2YWVJmrln64HHdbFnSD5pntampRZua2tY9x8KtjN46lqxBb6Nh3WDMOffza7WZ2bSZ3b88K3a/pJk1znF5+Z8zZvY9Sfsk+QZjy7NmL0rS4OCgu6+td/27iJBEMi7GZDXGxR/j4o9xWY0x8RfacWmTRs9PK7W9TcnNkm1Na2fnDT3UfrmoTf1vHUvq0cfXD9pQnHL3jL0s6YvLr78o6c9XvsHM2s1ss/da0qclxcq8LgAAQCiUG4z9W0mfMrOzkj61/LPMbJeZvbr8nm5JR8zstKQTkl5xzv1lmdcFAAAIhbI28DvnEpKe9Dl+WdL+5dcXJD1SznUAAADCigz8AABEVPzCjJIftCqTXtQHmtO9omRSEAjGAACIIK+G5eFTZxWfXpAe2aIz6pWU/fYoAVntlLtnDAAANDAvGayXnZ9ksLXHzBgAABG3VlFxksHWBsEYAADwLyq+fU4SRcWrjWAMAABIWr2PbLanWdrNPrJqIxgDAAB3OLB3UKPnp3Vhck6juqWZXR0a7hqXdhCQVQPBGAAAWGVlUfFYqkdStqh4U+anJVEOqVL4NiUAAPA1NNCthzZvUdvUoq6dXlDs8j6dud6rVCbNty0riJkxNJynnn5Y195vXXF0j7ZtXdQr330vkD4BQFh5+8hGz0/rjdNXNNvTrN57mzVxiwSxlcLMGBrO6kCs8HEAQPmGBrr1RPtO9U22S6m7NDLeoZMz4zpy9XjQXWt4zIwBAICiePvImhbn1TRB+otKIRgDAABFGxroViK5qLb3PiD9RYUQjAEAgA0j/UXlEIwBAICSFEp/QRml4hGMoeFs27rou1l/29bFAHoDANGWX0YpPr2gmPaxj2yDCMbQcPzSVySScd3X1htAbwAAfukvZnZ16MkHSX9RDIIxAABQEblZssk5JdOdGtFF9pEVgWAMAABUTP6yZVLL6S8W57Snk2XLtRCMAQCAivKWLQ+fOkv6iyIQjAEAgKog/UVxCMYAAEDVkP5ifQRjAACgqkh/URjBGAAAqDrSX6yNYAyh8tTTD6+ZENYvPxkAoLZIf7EawRhCxS8QK3QcAFB7pL+4E8EYAACoOdJf3EYwBgAAAuOX/iJq+8gIxgAAQKBWpr8YSS1qeFd00l8QjAEAgMBFOf0FwRhCZdvWxTW/TQkAqG9+6S+isI+MYAyhQvoKAGh8+ekvZtOtGum7Ger0FwRjAACg7kQp/QXBGAAAqEsr01/okS06o/AtWzYF3QEAAIBCDuwd1BPtO3XPj6TRsWaNjHdo4lZCEwvngu5aRTAzBgAA6l6Y018QjAEAgIZw5z6y1tCkvyAYAwAADePO9Bc3QpH+gmAMAAA0nDClvyAYQ9U89fTDayZgLZQP7GOfHpZkPi1OR1+PrXHePbnErqVcs5z+AgCC4Zf+QsruI2ukZUuCMVSNX2BT6PhtfoHY7eOlnHf9a5bTXwBAUFamv7i9j6xxli1JbQEAABpeI6e/YGYMAACEwqr0F7qo4a7xuk9/QTAGAABCw7eMUp2nvyAYAwAAodJo6S8IxlA127YurvntxMKc1vo2ZTHnLe2a5fQXAFCP8tNfjOqWZnZ11GX6C4IxVE2p6SCOvh7b8HkTybjua+st6XqFzgsAaGwr95HVY/oLgjEAABBq9Z7+gmAMAABEwoG9gxo9P63UjTb9RNIZ1cc+MoIxAAAQOX0tO2TapnltlXQ50L6Q9BUAACBABGMAAAABIhgDAAAIUFnBmJn9spm9a2YZM3u0wPs+a2Y/NrNzZvblcq4JAABQqqGBbsUvzOjU2xf19tik/uf19wOvYVnuBv6YpL8v6Q/WeoOZNUv6fUmfUvYrCyfN7GXn3GiZ10ad+/hnhuXc6uStZk7OSWsldt22Nb1m8tX1coE99fTDJX8WABAN3rcq85PBPvlgQlIw36osKxhzzo1Jkpnff1Rz9kk655y7sPze70j6nCSCsZDzC8QKHc8y32BK8s+sX+x7ivksACA6VhUVTy1qeNeJQIqK1yK1RY+kybyf45Lqt3Q6AACIhPyi4ncmg61tUfF1gzEz+ytJO32avuKc+/MirrF2kUH/6x2UdFCStm/frkQyXsQloiOdWWygMdlT8TOude+3x2XtazbOuFVOYz0vtcO4rMaY+GNc/IVpXLp7pG7drWSyVTcTi9LNLbrZtlsXmlOKN13Xpqa2qvdh3WDMOffzZV4jLumBvJ97VSC7mnPuRUkvStLg4KArt95g2FSiBmMjW+veixmXKI5b1J+XtTAuqzEm/hgXf6EclzZl95HdnFPy/k3K9F3WcNcl/a87Hqz6DFktlilPSho0s5+SdEnSFyT9Sg2uCwAAULT8Zcuk+hVL9WhmcU57Oqu7bFluaotfNLO4pI9JesXMXls+vsvMXpUk51xa0pckvSZpTNKfOufeLa/baARm/qvR2eNrrVQ7bdu66Nuy1vFi3lPMZwEAGBro1oG9g2qbWtS10wuaubpFZ673VjX9RbnfpvyepO/5HL8saX/ez69KerWca6Hx/M1rsZpfk/QVAIBKyKW/+NGcRh9ormr6CwqFAwAA+KhV+guCMQAAgDXUIv0FwRgAAEABQwPdGlK3Rs9P643TVzTb0yzt7lU2YUT5y5YEYwAAAEXIzZJNzmk23aqRvpsa7hqXdpQXkBGMAQAAFKlQ+otSlZXaAgAAIGrWSn9RKmbGoKeefti3kPa2rYvrpor4+GeGfQt/mzk5JxWuhrXxtqOvx/SxTw/7tO9RNkdZuuR7KWccAADRk5/+Yvb++0o+DzNj8A1ACh3P5xeI3T7u35Y9XmqbCraXcy/lfBYAEE1DA916aPMWtU2VnlycYAwAAKAM3rJlqQjGAAAAAkQwBgAAECCCMQAAgAARjEHbtvpvOlzreD4zV+C4f1v2eKltKthezr2U81kAAEpFaguUlbbhb16LVbAnxTn6+uprJpJx3ddWeo4XqbxxAACgVMyMAQAABIhgDAAAIEAEYwAAAAEiGAMAAAgQwRgAAECACMYAAAACRDAGAAAQIPKMoaCnnn5Y195vXXV829bFsvJyffwzw3LOVh33ksiu1RZEXjMAAKqJYAwF+QVihY4Xyy/YKnR8vTYAABoVy5QAAAABIhgDAAAIEMEYAABAgAjGAAAAAkQwhoK2bV3c0PFied+a9DteqA0AgLDh25QoqJz0FYWQogIAgCxmxgAAAAJEMAYAABAggjEAAIAAEYwBAAAEiGAMAAAgQARjAAAAASIYAwAACBDBGAAAQIAIxgAAAAJEMAYAABAggjEAAIAAEYwBAAAEiGAMAAAgQARjAAAAASIYAwAACBDBGAAAQIAIxgAAAAJEMAYAABAggjEAAIAAEYwBAAAEiGAMAAAgQARjAAAAASIYAwAACBDBGAAAQIDKCsbM7JfN7F0zy5jZowXed9HMzpjZKTN7q5xrAgAAhElLmZ+PSfr7kv6giPf+nHNutszrAQAAhEpZwZhzbkySzKwyvQEAAIiYWu0Zc5JeN7O3zexgja4JAABQ99adGTOzv5K006fpK865Py/yOj/rnLtsZjskfd/M3nPOvbHG9Q5K8gK2Dz/26f8lVuQ1oqJLEsu9qzEu/hgXf4zLaoyJP8bFH+Pi76dL+dC6wZhz7udLOfGKc1xe/ueMmX1P0j5JvsGYc+5FSS9Kkpm95Zxb84sBUcSY+GNc/DEu/hiX1RgTf4yLP8bFX6lfUqz6MqWZtZvZZu+1pE8ru/EfAAAg8spNbfGLZhaX9DFJr5jZa8vHd5nZq8tv65Z0xMxOSzoh6RXn3F+Wc10AAICwKPfblN+T9D2f45cl7V9+fUHSIyVe4sXSexdajIk/xsUf4+KPcVmNMfHHuPhjXPyVNC7mnKt0RwAAAFAkyiEBAAAEqG6CMTP7d2b2npm9Y2bfM7PONd73WTP7sZmdM7Mv17qftUbJKX8bGJeoPS/bzOz7ZnZ2+Z9b13hf6J+X9X73lvV7y+3vmNlHg+hnrRUxLp80sxvLz8YpM/vtIPpZS2b2TTObMTPfL5dF+FlZb1yi+Kw8YGZ/bWZjy/8N+mc+79n48+Kcq4s/yn7LsmX59XOSnvN5T7Ok85IekrRJ0mlJQ0H3vcrjslvZvCU/kPRogfddlNQVdH/raVwi+rw8L+nLy6+/7PfvURSel2J+98rua/0LSSbpcUnHg+53nYzLJyUdDrqvNR6XJyR9VFJsjfbIPStFjksUn5X7JX10+fVmST+pxN8tdTMz5px73TmXXv7xmKRen7ftk3TOOXfBOZeS9B1Jn6tVH4PgnBtzzv046H7UmyLHJXLPi7L3963l19+S9PkA+xKkYn73n5P0bZd1TFKnmd1f647WWBT/nViXyyYhv1bgLVF8VooZl8hxzk055364/PqmpDFJPSvetuHnpW6CsRX+sbJR5Uo9kibzfo5r9SBEFSWnVovi89LtnJuSsn9pSNqxxvvC/rwU87uP4vNR7D1/zMxOm9lfmNlHatO1uhbFZ6VYkX1WzKxf0t+WdHxF04afl7JSW2xUMaWVzOwrktKSDvmdwudYw38dtNYlpxpFBcYlcs/LBk4TuudlhWJ+96F8PtZRzD3/UNKDzrl5M9sv6b9KGqx6z+pbFJ+VYkT2WTGzeyX9F0n/3Dk3t7LZ5yMFn5eaBmNundJKZvZFSQckPemWF15XiEt6IO/nXkmXK9fDYKw3LkWeo+iSU42iAuMSuefFzKbN7H7n3NTytPjMGucI3fOyQjG/+1A+H+tY957z/8PinHvVzL5uZl3OuSjXIYzis7KuqD4rZtaqbCB2yDn3Zz5v2fDzUjfLlGb2WUn/UtLfc84trPG2k5IGzeynzGyTpC9IerlWfaxXRsmptUTxeXlZ0heXX39R0qoZxIg8L8X87l+W9KvL33x6XNINb4k3xNYdFzPbaWa2/Hqfsv+dSNS8p/Ulis/KuqL4rCzf7x9KGnPO/fs13rbh56WmM2Pr+Jqku5RdMpGkY8653zCzXZL+o3Nuv3MubWZfkvSast8K+qZz7t3gulx9ZvaLkv6DpO3Klpw65Zz7TP64KFty6nvL49Yi6U9cyEtOFTMuUXxeJP1bSX9qZv9E0oSkX5ayJcoUoedlrd+9mf3Gcvs3JL2q7LeezklakPRrQfW3Voocl1+S9Jtmlpb0gaQvrLFSERpm9pKy3wzssmyJv9+R1CpF91mRihqXyD0rkn5W0j+SdMbMTi0f+1eS+qTSnxcy8AMAAASobpYpAQAAoohgDAAAIEAEYwAAAAEiGAMAAAgQwRgAAECACMYAAAACRDAGAAAQIIIxAACAAP3/t4WKlaFf0PEAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 720x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "iris = datasets.load_iris()\n",
    "\n",
    "X, y = get_linearly_separable_iris_data()\n",
    "\n",
    "axes = [-2, 2, -1.5, 2]\n",
    "\n",
    "# C big to force hard-margin SVM\n",
    "linear_clf = LinearSVC(loss='hinge', C=1000000)\n",
    "linear_clf.fit(X, y)\n",
    "\n",
    "plt.figure(figsize=(10, 8))\n",
    "plot_decision_boundary(linear_clf, axes)\n",
    "plot_dataset(X, y, axes)\n",
    "\n",
    "b = linear_clf.intercept_[0]\n",
    "w = linear_clf.coef_[0]\n",
    "\n",
    "# picks a random point to try\n",
    "xi, yi = X[51], y[51]\n",
    "\n",
    "# highlights the chosen point\n",
    "plt.scatter([xi[0]], [xi[1]], c='orange', alpha=0.7, s=280)\n",
    "\n",
    "ti = 1 if yi == 0 else -1\n",
    "gamma = ti * (w.T @ xi + b)\n",
    "xi_line = xi - gamma * w / np.linalg.norm(w)\n",
    "\n",
    "plt.scatter(xi_line[0], xi_line[1], c='k') \n",
    "print(xi_line) # [2.18464802 2.27591335] far from line"
   ]
  }
 ],
 "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": "3.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}