22-SMO算法及原生代码实现（机器学习实战源码）.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "难，有些还没有理解，所以只给出结论"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<div  align=\"center\">    \n",
    "    <img src = './img/28.jpg', width=600, height=200/>\n",
    "    <img src = './img/29.jpg', width=600, height=200/>\n",
    "</div>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "# coding=utf-8\n",
    "#svm原生代码实现，使用最简单的方法实现SMO\n",
    "from numpy import *\n",
    "\n",
    "import matplotlib as mpl\n",
    "import matplotlib.pyplot as plt\n",
    "from matplotlib.patches import Circle\n",
    "# 数据的加载\n",
    "def loadDataSet(fileName):\n",
    "    dataMat = []; labelMat = []\n",
    "    fr = open(fileName)\n",
    "    for line in fr.readlines():\n",
    "        lineArr = line.strip().split('\\t')\n",
    "        dataMat.append([float(lineArr[0]), float(lineArr[1])])\n",
    "        labelMat.append(float(lineArr[2]))\n",
    "    return dataMat,labelMat\n",
    "\n",
    "#随机选择一个不等于i值的j （例如选择alpha1 和 alpha2）\n",
    "def selectJrand(i,m):\n",
    "    j=i\n",
    "    while (j==i):\n",
    "        j = int(random.uniform(0,m))\n",
    "    return j\n",
    "\n",
    "#进行剪辑\n",
    "def clipAlpha(aj,H,L):\n",
    "    if aj > H:\n",
    "        aj = H\n",
    "    if L > aj:\n",
    "        aj = L\n",
    "    return aj\n",
    "\n",
    "def smoSimple(dataMatIn, classLabels, C, toler, maxIter):\n",
    "    dataMatrix = mat(dataMatIn); labelMat = mat(classLabels).transpose()\n",
    "    b = 0; m,n = shape(dataMatrix)\n",
    "    alphas = mat(zeros((m,1)))\n",
    "    iter = 0\n",
    "    while (iter < maxIter):\n",
    "        alphaPairsChanged = 0\n",
    "        for i in range(m):\n",
    "            gXi = float(multiply(alphas,labelMat).T*(dataMatrix*dataMatrix[i,:].T)) + b\n",
    "            Ei = gXi - float(labelMat[i])\n",
    "            if ((labelMat[i]*Ei < -toler) and (alphas[i] < C)) or ((labelMat[i]*Ei > toler) and (alphas[i] > 0)):\n",
    "                j = selectJrand(i,m)\n",
    "                gXj = float(multiply(alphas,labelMat).T*(dataMatrix*dataMatrix[j,:].T)) + b\n",
    "                Ej = gXj - float(labelMat[j])\n",
    "                alphaIold = alphas[i].copy(); alphaJold = alphas[j].copy();\n",
    "                if (labelMat[i] != labelMat[j]):\n",
    "                    L = max(0, alphas[j] - alphas[i])\n",
    "                    H = min(C, C + alphas[j] - alphas[i])\n",
    "                else:\n",
    "                    L = max(0, alphas[j] + alphas[i] - C)\n",
    "                    H = min(C, alphas[j] + alphas[i])\n",
    "                if L==H: print(\"L==H\"); continue\n",
    "                eta = 2.0 * dataMatrix[i,:]*dataMatrix[j,:].T - dataMatrix[i,:]*dataMatrix[i,:].T - dataMatrix[j,:]*dataMatrix[j,:].T\n",
    "                if eta >= 0: print(\"eta>=0\"); continue\n",
    "                alphas[j] -= labelMat[j]*(Ei - Ej)/eta\n",
    "                alphas[j] = clipAlpha(alphas[j],H,L)\n",
    "                if (abs(alphas[j] - alphaJold) < 0.00001): print(\"j not moving enough\"); continue\n",
    "                alphas[i] += labelMat[j]*labelMat[i]*(alphaJold - alphas[j])\n",
    "\n",
    "                b1 = b - Ei- labelMat[i]*(alphas[i]-alphaIold)*dataMatrix[i,:]*dataMatrix[i,:].T - labelMat[j]*(alphas[j]-alphaJold)*dataMatrix[i,:]*dataMatrix[j,:].T\n",
    "                b2 = b - Ej- labelMat[i]*(alphas[i]-alphaIold)*dataMatrix[i,:]*dataMatrix[j,:].T - labelMat[j]*(alphas[j]-alphaJold)*dataMatrix[j,:]*dataMatrix[j,:].T\n",
    "                if (0 < alphas[i]) and (C > alphas[i]): b = b1\n",
    "                elif (0 < alphas[j]) and (C > alphas[j]): b = b2\n",
    "                else: b = (b1 + b2)/2.0\n",
    "                alphaPairsChanged += 1\n",
    "                print(\"iter: %d i:%d, pairs changed %d\" % (iter,i,alphaPairsChanged))\n",
    "        if (alphaPairsChanged == 0): iter += 1\n",
    "        else: iter = 0\n",
    "        print(\"iteration number: %d\" % iter)\n",
    "    return b,alphas"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 计算我们的W\n",
    "def calcWs(alphas,dataArr,classLabels):\n",
    "    X = mat(dataArr); labelMat = mat(classLabels).transpose()\n",
    "    m,n = shape(X)\n",
    "    w = zeros((n,1))\n",
    "    for i in range(m):\n",
    "        w += multiply(alphas[i]*labelMat[i],X[i,:].T)\n",
    "    return w"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "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": [
    "dataMat,labelMat = loadDataSet('./files/testSet.txt')\n",
    "#画出原始的数据图像\n",
    "fig = plt.figure()\n",
    "ax = fig.add_subplot(111)\n",
    "cm_dark = mpl.colors.ListedColormap(['g', 'r'])\n",
    "ax.scatter(array(dataMat)[:,0],array(dataMat)[:,1],c=array(labelMat).squeeze(),cmap=cm_dark,s=30)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "iter: 0 i:0, pairs changed 1\n",
      "iter: 0 i:2, pairs changed 2\n",
      "L==H\n",
      "j not moving enough\n",
      "L==H\n",
      "L==H\n",
      "L==H\n",
      "L==H\n",
      "L==H\n",
      "iter: 0 i:23, pairs changed 3\n",
      "L==H\n",
      "j not moving enough\n",
      "iter: 0 i:26, pairs changed 4\n",
      "iter: 0 i:29, pairs changed 5\n",
      "j not moving enough\n",
      "L==H\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "L==H\n",
      "j not moving enough\n",
      "iter: 0 i:50, pairs changed 6\n",
      "iter: 0 i:51, pairs changed 7\n",
      "iter: 0 i:54, pairs changed 8\n",
      "j not moving enough\n",
      "iter: 0 i:80, pairs changed 9\n",
      "j not moving enough\n",
      "iteration number: 0\n",
      "j not moving enough\n",
      "iter: 0 i:2, pairs changed 1\n",
      "j not moving enough\n",
      "L==H\n",
      "j not moving enough\n",
      "iter: 0 i:10, pairs changed 2\n",
      "iter: 0 i:11, pairs changed 3\n",
      "j not moving enough\n",
      "L==H\n",
      "j not moving enough\n",
      "L==H\n",
      "j not moving enough\n",
      "iter: 0 i:29, pairs changed 4\n",
      "iter: 0 i:31, pairs changed 5\n",
      "j not moving enough\n",
      "iter: 0 i:50, pairs changed 6\n",
      "iter: 0 i:51, pairs changed 7\n",
      "L==H\n",
      "iter: 0 i:54, pairs changed 8\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 0\n",
      "iter: 0 i:0, pairs changed 1\n",
      "iter: 0 i:8, pairs changed 2\n",
      "j not moving enough\n",
      "iter: 0 i:16, pairs changed 3\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 0\n",
      "j not moving enough\n",
      "L==H\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "L==H\n",
      "L==H\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "L==H\n",
      "j not moving enough\n",
      "L==H\n",
      "iter: 0 i:46, pairs changed 1\n",
      "iter: 0 i:54, pairs changed 2\n",
      "iter: 0 i:55, pairs changed 3\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "L==H\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 0\n",
      "L==H\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "L==H\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "L==H\n",
      "L==H\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 1\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "L==H\n",
      "j not moving enough\n",
      "L==H\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 2\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "L==H\n",
      "iter: 2 i:46, pairs changed 1\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 0\n",
      "iter: 0 i:0, pairs changed 1\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 0\n",
      "L==H\n",
      "j not moving enough\n",
      "L==H\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iter: 0 i:29, pairs changed 1\n",
      "j not moving enough\n",
      "iter: 0 i:52, pairs changed 2\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iter: 0 i:69, pairs changed 3\n",
      "j not moving enough\n",
      "iteration number: 0\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "L==H\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 1\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "L==H\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iter: 1 i:46, pairs changed 1\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 0\n",
      "iter: 0 i:0, pairs changed 1\n",
      "j not moving enough\n",
      "iter: 0 i:10, pairs changed 2\n",
      "iter: 0 i:17, pairs changed 3\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 0\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 1\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iter: 1 i:29, pairs changed 1\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 0\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iter: 0 i:26, pairs changed 1\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iter: 0 i:55, pairs changed 2\n",
      "j not moving enough\n",
      "L==H\n",
      "L==H\n",
      "j not moving enough\n",
      "iteration number: 0\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "L==H\n",
      "j not moving enough\n",
      "iter: 0 i:29, pairs changed 1\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 0\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "L==H\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 1\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 2\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "L==H\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 3\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iter: 3 i:54, pairs changed 1\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iter: 3 i:96, pairs changed 2\n",
      "iter: 3 i:97, pairs changed 3\n",
      "j not moving enough\n",
      "iteration number: 0\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "L==H\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "L==H\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iter: 0 i:55, pairs changed 1\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iter: 0 i:97, pairs changed 2\n",
      "iteration number: 0\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "L==H\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iter: 0 i:52, pairs changed 1\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 0\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "L==H\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 1\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "L==H\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 2\n",
      "iter: 2 i:0, pairs changed 1\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 0\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 1\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iter: 1 i:29, pairs changed 1\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iter: 1 i:96, pairs changed 2\n",
      "j not moving enough\n",
      "iteration number: 0\n",
      "L==H\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 1\n",
      "L==H\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "L==H\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 2\n",
      "L==H\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iter: 2 i:23, pairs changed 1\n",
      "L==H\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iter: 2 i:29, pairs changed 2\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 0\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 1\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 2\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iter: 2 i:43, pairs changed 1\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 0\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "L==H\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iter: 0 i:54, pairs changed 1\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 0\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "L==H\n",
      "j not moving enough\n",
      "iter: 0 i:23, pairs changed 1\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 0\n",
      "L==H\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 1\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iter: 1 i:29, pairs changed 1\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iter: 1 i:54, pairs changed 2\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 0\n",
      "L==H\n",
      "L==H\n",
      "L==H\n",
      "j not moving enough\n",
      "L==H\n",
      "iter: 0 i:24, pairs changed 1\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 0\n",
      "j not moving enough\n",
      "iter: 0 i:17, pairs changed 1\n",
      "iter: 0 i:24, pairs changed 2\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iter: 0 i:43, pairs changed 3\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 0\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 1\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 2\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iter: 2 i:69, pairs changed 1\n",
      "iter: 2 i:96, pairs changed 2\n",
      "iteration number: 0\n",
      "L==H\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iter: 0 i:55, pairs changed 1\n",
      "j not moving enough\n",
      "iteration number: 0\n",
      "j not moving enough\n",
      "iter: 0 i:8, pairs changed 1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 0\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 1\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 2\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iter: 2 i:29, pairs changed 1\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 0\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 1\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 2\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 3\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 4\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 5\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 6\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 7\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 8\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 9\n",
      "j not moving enough\n",
      "iter: 9 i:23, pairs changed 1\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 0\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "L==H\n",
      "j not moving enough\n",
      "iter: 0 i:55, pairs changed 1\n",
      "j not moving enough\n",
      "iteration number: 0\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 1\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 2\n",
      "j not moving enough\n",
      "iter: 2 i:17, pairs changed 1\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 0\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 1\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 2\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 3\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iter: 3 i:29, pairs changed 1\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 0\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 1\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 2\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 3\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 4\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 5\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 6\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 7\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 8\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 9\n",
      "iter: 9 i:17, pairs changed 1\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 0\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 1\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 2\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 3\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 4\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 5\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 6\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 7\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 8\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 9\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 10\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 11\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 12\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 13\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 14\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 15\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 16\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 17\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 18\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 19\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 20\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 21\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iter: 21 i:55, pairs changed 1\n",
      "iteration number: 0\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 1\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 2\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 3\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 4\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 5\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 6\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 7\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 8\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 9\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 10\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 11\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 12\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 13\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 14\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 15\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 16\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 17\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 18\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 19\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 20\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 21\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 22\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 23\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 24\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 25\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 26\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 27\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 28\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 29\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 30\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 31\n",
      "j not moving enough\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 32\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 33\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 34\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 35\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 36\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 37\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 38\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 39\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "j not moving enough\n",
      "iteration number: 40\n",
      "b= [[-3.87540774]]\n",
      "alphas= [[0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [5.30352291e-02]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [3.07411288e-01]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [6.93889390e-18]\n",
      " [0.00000000e+00]\n",
      " [8.06576343e-02]\n",
      " [2.79788883e-01]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]\n",
      " [0.00000000e+00]]\n",
      "w= [[ 0.81790395]\n",
      " [-0.23375555]]\n"
     ]
    }
   ],
   "source": [
    "b,alphas = smoSimple(dataMat,labelMat,0.6,0.001,40)\n",
    "w = calcWs(alphas,dataMat,labelMat)\n",
    "print('b=',b)\n",
    "print('alphas=',alphas)\n",
    "print('w=',w)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "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": [
    "#画决策平面\n",
    "fig = plt.figure()\n",
    "ax = fig.add_subplot(111)\n",
    "cm_dark = mpl.colors.ListedColormap(['g', 'r'])\n",
    "ax.scatter(array(dataMat)[:,0],array(dataMat)[:,1],c=array(labelMat).squeeze(),cmap=cm_dark,s=30)\n",
    "x = arange(-2.0, 12.0, 0.1)\n",
    "y = (-w[0]*x - b)/w[1]  # w[0]x1 + w[1]x2 + b = 0 ==> 可以得出x2的值，从而在二维中画出\n",
    "ax.plot(x,y.reshape(-1,1))\n",
    "ax.axis([-2,12,-8,6])\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "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": [
    "#画支持向量\n",
    "fig = plt.figure()\n",
    "ax = fig.add_subplot(111)\n",
    "cm_dark = mpl.colors.ListedColormap(['g', 'r'])\n",
    "ax.scatter(array(dataMat)[:,0],array(dataMat)[:,1],c=array(labelMat).squeeze(),cmap=cm_dark,s=30)\n",
    "alphas_non_zeros_index = where(alphas>0)  # appha > 0 说明是我们的支撑点\n",
    "for i in alphas_non_zeros_index[0]:\n",
    "    circle = Circle((dataMat[i][0], dataMat[i][1]), 0.2, facecolor='none', edgecolor=(0,0.8,0.8), linewidth=3, alpha=0.5)\n",
    "    ax.add_patch(circle)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "END"
   ]
  }
 ],
 "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": "3.7.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}