Week5_SVM_Assignment.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "9ad88f3e",
   "metadata": {},
   "source": [
    "# Multiclass SVM 구현"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "1ee53277",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0         setosa\n",
      "1         setosa\n",
      "2         setosa\n",
      "3         setosa\n",
      "4         setosa\n",
      "         ...    \n",
      "145    virginica\n",
      "146    virginica\n",
      "147    virginica\n",
      "148    virginica\n",
      "149    virginica\n",
      "Name: species, Length: 150, dtype: object\n"
     ]
    }
   ],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "from sklearn.svm import SVC\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.pipeline import Pipeline\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "from sklearn.metrics import accuracy_score\n",
    "\n",
    "#IRIS 데이터 로드\n",
    "iris =  sns.load_dataset('iris') \n",
    "X= iris.iloc[:,:4] #학습할데이터\n",
    "y = iris.iloc[:,-1] #타겟\n",
    "print(y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "57c67bc8",
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=48)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "88a4203d",
   "metadata": {},
   "outputs": [],
   "source": [
    "def standardization(train, test):\n",
    "    scaler = StandardScaler()\n",
    "    train = scaler.fit_transform(train)\n",
    "    test = scaler.transform(test)\n",
    "    return train, test\n",
    "\n",
    "X_train, X_test = standardization(X_train, X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "403590f5",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[ 0.78522493,  0.32015325,  0.77221097,  1.04726529],\n",
       "       [-0.26563371, -1.29989934,  0.0982814 , -0.11996537],\n",
       "       [ 0.43493872,  0.78302542,  0.94069336,  1.43634218],\n",
       "       [-0.84944407,  0.78302542, -1.24957775, -1.28719604],\n",
       "       [-0.38239578, -1.7627715 ,  0.15444219,  0.13941922],\n",
       "       [ 0.55170079, -0.374155  ,  1.05301496,  0.7878807 ],\n",
       "       [ 0.31817664, -0.14271892,  0.65988937,  0.7878807 ],\n",
       "       [ 0.20141457, -0.374155  ,  0.43524618,  0.39880381],\n",
       "       [-1.66677857, -0.14271892, -1.36189934, -1.28719604],\n",
       "       [-0.14887164, -0.60559109,  0.21060299,  0.13941922],\n",
       "       [-0.14887164, -1.06846325, -0.12636179, -0.24965767],\n",
       "       [ 0.31817664, -0.60559109,  0.15444219,  0.13941922],\n",
       "       [ 0.66846286, -0.83702717,  0.88453256,  0.91757299],\n",
       "       [ 0.0846525 , -0.14271892,  0.77221097,  0.7878807 ],\n",
       "       [-0.49915786, -0.14271892,  0.43524618,  0.39880381],\n",
       "       [-0.26563371, -0.60559109,  0.65988937,  1.04726529],\n",
       "       [ 2.18636979,  1.70876975,  1.67078373,  1.30664988],\n",
       "       [-0.49915786,  1.94020584, -1.13725615, -1.02781144],\n",
       "       [-0.96620614,  1.0144615 , -1.36189934, -1.15750374],\n",
       "       [ 1.60255943,  0.32015325,  1.27765815,  0.7878807 ],\n",
       "       [-0.84944407,  1.70876975, -1.02493456, -1.02781144],\n",
       "       [ 0.901987  , -0.14271892,  0.37908538,  0.26911151],\n",
       "       [-1.19973028,  0.78302542, -1.02493456, -1.28719604],\n",
       "       [-0.61591993,  1.47733367, -1.24957775, -1.28719604],\n",
       "       [-0.49915786,  1.47733367, -1.24957775, -1.28719604],\n",
       "       [-0.03210957, -0.83702717,  0.77221097,  0.91757299],\n",
       "       [-1.31649236,  0.32015325, -1.19341695, -1.28719604],\n",
       "       [ 1.25227322,  0.08871717,  0.94069336,  1.17695758],\n",
       "       [-1.08296821,  0.08871717, -1.24957775, -1.41688833],\n",
       "       [-1.08296821, -1.29989934,  0.43524618,  0.6581884 ],\n",
       "       [ 0.43493872, -0.374155  ,  0.32292459,  0.13941922],\n",
       "       [ 1.60255943,  1.24589758,  1.33381894,  1.69572677],\n",
       "       [ 0.31817664, -0.374155  ,  0.54756778,  0.26911151],\n",
       "       [ 1.83608357, -0.60559109,  1.33381894,  0.91757299],\n",
       "       [-0.03210957, -0.83702717,  0.0982814 ,  0.00972692],\n",
       "       [ 0.20141457, -1.99420759,  0.71605017,  0.39880381],\n",
       "       [ 1.01874907,  0.08871717,  1.05301496,  1.56603447],\n",
       "       [-0.84944407,  1.0144615 , -1.30573855, -1.15750374],\n",
       "       [-0.14887164, -0.14271892,  0.26676379,  0.00972692],\n",
       "       [-0.38239578, -1.53133542,  0.0421206 , -0.11996537],\n",
       "       [ 0.55170079, -1.29989934,  0.71605017,  0.91757299],\n",
       "       [ 1.48579736, -0.14271892,  1.22149735,  1.17695758],\n",
       "       [-0.38239578, -1.06846325,  0.37908538,  0.00972692],\n",
       "       [-0.96620614, -1.7627715 , -0.23868339, -0.24965767],\n",
       "       [ 1.25227322,  0.08871717,  0.77221097,  1.43634218],\n",
       "       [ 0.66846286, -0.60559109,  1.05301496,  1.17695758],\n",
       "       [ 0.43493872, -1.99420759,  0.43524618,  0.39880381],\n",
       "       [-0.14887164, -0.60559109,  0.43524618,  0.13941922],\n",
       "       [-0.732682  , -0.83702717,  0.0982814 ,  0.26911151],\n",
       "       [ 2.06960772, -0.14271892,  1.61462293,  1.17695758],\n",
       "       [ 1.13551115, -0.60559109,  0.60372857,  0.26911151],\n",
       "       [-0.84944407, -1.29989934, -0.40716578, -0.11996537],\n",
       "       [-0.14887164, -1.29989934,  0.71605017,  1.04726529],\n",
       "       [ 0.66846286, -0.374155  ,  0.32292459,  0.13941922],\n",
       "       [ 1.01874907,  0.08871717,  0.37908538,  0.26911151],\n",
       "       [-0.26563371, -0.83702717,  0.26676379,  0.13941922],\n",
       "       [ 1.25227322,  0.08871717,  0.65988937,  0.39880381],\n",
       "       [-0.84944407,  0.55158933, -1.13725615, -0.89811915],\n",
       "       [-1.08296821, -0.14271892, -1.30573855, -1.28719604],\n",
       "       [ 0.20141457, -0.83702717,  0.77221097,  0.52849611],\n",
       "       [-1.08296821, -1.53133542, -0.23868339, -0.24965767],\n",
       "       [-0.03210957,  2.17164192, -1.41806014, -1.28719604],\n",
       "       [-0.96620614,  1.0144615 , -1.19341695, -0.76842685],\n",
       "       [ 0.55170079, -0.83702717,  0.65988937,  0.7878807 ],\n",
       "       [ 1.7193215 , -0.374155  ,  1.44614054,  0.7878807 ],\n",
       "       [-0.732682  ,  2.403078  , -1.24957775, -1.41688833],\n",
       "       [-0.14887164,  3.09738625, -1.24957775, -1.02781144],\n",
       "       [ 1.13551115, -0.14271892,  0.99685416,  1.17695758],\n",
       "       [-1.43325443,  0.08871717, -1.24957775, -1.28719604],\n",
       "       [-0.03210957, -0.83702717,  0.21060299, -0.24965767],\n",
       "       [ 1.01874907, -0.14271892,  0.82837177,  1.43634218],\n",
       "       [ 1.36903529,  0.32015325,  0.54756778,  0.26911151],\n",
       "       [-0.96620614,  0.32015325, -1.41806014, -1.28719604],\n",
       "       [-0.96620614,  0.78302542, -1.24957775, -1.28719604],\n",
       "       [ 0.55170079,  0.55158933,  1.27765815,  1.69572677],\n",
       "       [ 2.18636979, -0.60559109,  1.67078373,  1.04726529],\n",
       "       [ 0.0846525 , -0.14271892,  0.26676379,  0.39880381],\n",
       "       [-1.31649236,  0.32015325, -1.36189934, -1.28719604],\n",
       "       [ 0.66846286,  0.32015325,  0.88453256,  1.43634218],\n",
       "       [-0.96620614,  0.78302542, -1.19341695, -1.02781144],\n",
       "       [ 0.20141457,  0.78302542,  0.43524618,  0.52849611],\n",
       "       [-1.19973028, -0.14271892, -1.30573855, -1.41688833],\n",
       "       [ 1.01874907,  0.55158933,  1.10917575,  1.17695758],\n",
       "       [ 0.78522493, -0.60559109,  0.49140698,  0.39880381],\n",
       "       [-1.43325443,  1.24589758, -1.53038174, -1.28719604],\n",
       "       [-0.03210957, -0.60559109,  0.77221097,  1.56603447],\n",
       "       [-1.78354064, -0.14271892, -1.47422094, -1.41688833],\n",
       "       [ 1.01874907, -1.29989934,  1.16533655,  0.7878807 ],\n",
       "       [-1.08296821,  1.24589758, -1.30573855, -1.41688833],\n",
       "       [-0.49915786,  1.94020584, -1.36189934, -1.02781144],\n",
       "       [-1.43325443,  0.32015325, -1.30573855, -1.28719604],\n",
       "       [-0.03210957, -0.83702717,  0.77221097,  0.91757299],\n",
       "       [ 0.55170079, -0.60559109,  0.77221097,  0.39880381],\n",
       "       [-0.96620614,  0.55158933, -1.30573855, -1.28719604],\n",
       "       [ 2.41989393,  1.70876975,  1.50230134,  1.04726529],\n",
       "       [-1.66677857,  0.32015325, -1.36189934, -1.28719604],\n",
       "       [-1.19973028, -0.14271892, -1.30573855, -1.15750374],\n",
       "       [ 1.01874907,  0.55158933,  1.10917575,  1.69572677],\n",
       "       [-0.96620614, -2.45707976, -0.12636179, -0.24965767],\n",
       "       [-0.732682  ,  1.0144615 , -1.24957775, -1.28719604],\n",
       "       [ 2.18636979, -1.06846325,  1.78310533,  1.43634218],\n",
       "       [ 1.13551115,  0.32015325,  1.22149735,  1.43634218],\n",
       "       [-0.96620614, -0.14271892, -1.19341695, -1.28719604],\n",
       "       [-0.84944407,  1.47733367, -1.24957775, -1.02781144],\n",
       "       [ 2.18636979, -0.14271892,  1.33381894,  1.43634218],\n",
       "       [-0.14887164,  1.70876975, -1.13725615, -1.15750374],\n",
       "       [-0.26563371, -0.14271892,  0.43524618,  0.39880381],\n",
       "       [-1.19973028,  0.08871717, -1.19341695, -1.28719604],\n",
       "       [ 0.78522493, -0.14271892,  1.16533655,  1.30664988],\n",
       "       [ 0.901987  , -0.374155  ,  0.49140698,  0.13941922],\n",
       "       [-0.732682  ,  0.78302542, -1.30573855, -1.28719604],\n",
       "       [-0.84944407,  1.70876975, -1.19341695, -1.28719604],\n",
       "       [-1.43325443,  0.78302542, -1.30573855, -1.15750374],\n",
       "       [-0.03210957, -1.06846325,  0.15444219,  0.00972692],\n",
       "       [-1.08296821,  0.08871717, -1.24957775, -1.28719604],\n",
       "       [-0.26563371, -0.374155  , -0.070201  ,  0.13941922],\n",
       "       [ 0.31817664, -0.14271892,  0.49140698,  0.26911151],\n",
       "       [-0.38239578, -1.53133542, -0.0140402 , -0.24965767],\n",
       "       [ 0.66846286,  0.32015325,  0.43524618,  0.39880381],\n",
       "       [-0.84944407,  1.0144615 , -1.30573855, -1.28719604]])"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "9861d8a6",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[-0.14887164, -0.374155  ,  0.26676379,  0.13941922],\n",
       "       [ 0.31817664, -0.60559109,  0.54756778,  0.00972692],\n",
       "       [ 0.31817664, -1.06846325,  1.05301496,  0.26911151],\n",
       "       [-1.5500165 , -1.7627715 , -1.36189934, -1.15750374],\n",
       "       [ 0.0846525 ,  0.32015325,  0.60372857,  0.7878807 ],\n",
       "       [ 0.78522493, -0.14271892,  0.99685416,  0.7878807 ],\n",
       "       [-0.84944407,  1.70876975, -1.24957775, -1.15750374],\n",
       "       [ 0.20141457, -0.14271892,  0.60372857,  0.7878807 ],\n",
       "       [-0.38239578,  2.63451409, -1.30573855, -1.28719604],\n",
       "       [-0.38239578, -1.29989934,  0.15444219,  0.13941922],\n",
       "       [ 0.66846286,  0.08871717,  0.99685416,  0.7878807 ],\n",
       "       [-0.38239578,  1.0144615 , -1.36189934, -1.28719604],\n",
       "       [-0.49915786,  0.78302542, -1.13725615, -1.28719604],\n",
       "       [ 0.43493872, -0.60559109,  0.60372857,  0.7878807 ],\n",
       "       [ 0.55170079, -1.7627715 ,  0.37908538,  0.13941922],\n",
       "       [ 0.55170079,  0.55158933,  0.54756778,  0.52849611],\n",
       "       [-1.19973028,  0.78302542, -1.19341695, -1.28719604],\n",
       "       [ 0.20141457, -1.99420759,  0.15444219, -0.24965767],\n",
       "       [ 0.55170079, -1.29989934,  0.65988937,  0.39880381],\n",
       "       [ 1.25227322,  0.32015325,  1.10917575,  1.43634218],\n",
       "       [-1.66677857, -0.374155  , -1.30573855, -1.28719604],\n",
       "       [ 0.78522493, -0.14271892,  0.82837177,  1.04726529],\n",
       "       [ 1.01874907, -0.14271892,  0.71605017,  0.6581884 ],\n",
       "       [ 1.01874907,  0.08871717,  0.54756778,  0.39880381],\n",
       "       [ 1.60255943, -0.14271892,  1.16533655,  0.52849611],\n",
       "       [-0.96620614,  1.24589758, -1.30573855, -1.28719604],\n",
       "       [-0.49915786,  0.78302542, -1.24957775, -1.02781144],\n",
       "       [ 0.55170079,  0.78302542,  1.05301496,  1.56603447],\n",
       "       [ 0.66846286, -0.60559109,  1.05301496,  1.30664988],\n",
       "       [-0.26563371, -0.14271892,  0.21060299,  0.13941922]])"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "11719030",
   "metadata": {},
   "outputs": [],
   "source": [
    "# One VS Rest 구현을 위해 1) class가 0 or not 2)class가 1 or not을 구분하기 위한 머신 등을 미리 만들어주자\n",
    "svm_1 = SVC(kernel ='rbf', C = 5, gamma = 5)\n",
    "svm_2 = SVC(kernel ='rbf', C = 5, gamma = 5)\n",
    "svm_3 = SVC(kernel ='rbf', C = 5, gamma = 5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "f44f1f10",
   "metadata": {},
   "outputs": [],
   "source": [
    "y_train = pd.get_dummies(y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "f5f3730c",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0 0 0 0 0 0 1 0 1 0 0 1 1 0 0 0 1 0 0 0 1 0 0 0 0 1 1 0 0 0]\n",
      "[-1.12359969 -0.86782512 -0.65599247 -0.50194294 -0.76541147 -0.8819188\n",
      "  1.07735938 -0.99156769  0.50201986 -0.9984315  -0.84532712  0.17062549\n",
      "  0.34917127 -0.9813287  -0.72783399 -0.93313988  1.28153212 -0.56827872\n",
      " -0.73092732 -0.99670034  0.43553308 -0.96967771 -0.83939495 -1.03305682\n",
      " -0.75566609  1.13888006  0.42965012 -1.04268452 -0.93608147 -1.06090982]\n"
     ]
    }
   ],
   "source": [
    "svm_1.fit(X_train,y_train.iloc[:,0]) # 데이터 클레스가 0 or not 구분해주는 머신\n",
    "svm_2.fit(X_train,y_train.iloc[:,1]) # 데이터 클레스가 1 or not 구분해주는 머신\n",
    "svm_3.fit(X_train,y_train.iloc[:,2]) # 데이터 클레스가 2 or not 구분해주는 머신\n",
    "print(svm_1.predict(X_test)) #데이터 클래스가 0 or not을 구분해주는 애를 통해서 테스트 데이터의 클래스가 0인지, 0이 아닌인지 예측해보자\n",
    "\n",
    "print(svm_1.decision_function(X_test)) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "560043c6",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "3\n",
      "17\n",
      "18\n"
     ]
    }
   ],
   "source": [
    "# 부호가 모든 같은 경우가 있는가? > 모두 동점인 경우가 생길 것\n",
    "for i in range(len(X_test)):\n",
    "    # ~. decision_function을 이용하면 해당 데이터가 하이퍼플레인으로부터 얼마나 떨어져있는지 '거리'가 나온다!\n",
    "    # 다음은 그 값의 부호를 이용해 모두가 동점인 경우가 있는지 출력하는 함수 \n",
    "    if (np.sign(svm_1.decision_function(X_test)[i]) == np.sign(svm_2.decision_function(X_test)[i]))\\\n",
    "    and (np.sign(svm_2.decision_function(X_test)[i]) == np.sign(svm_3.decision_function(X_test)[i])):\n",
    "        print(i)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "ba9af14d",
   "metadata": {},
   "outputs": [],
   "source": [
    "result1 = svm_1.decision_function(X_test)   #setosa\n",
    "result2 = svm_2.decision_function(X_test)   #versicolor\n",
    "result3 = svm_3.decision_function(X_test)   #virginica"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "a00432f5",
   "metadata": {},
   "outputs": [],
   "source": [
    "result = [] #결과값 저장할 list\n",
    "for i in range(X_test.shape[0]):\n",
    "    if result1[i]==max(result1[i], result2[i], result3[i]):\n",
    "        result.append('setosa')\n",
    "    elif result2[i]==max(result1[i], result2[i], result3[i]):\n",
    "        result.append('versicolor')\n",
    "    elif result3[i]==max(result1[i], result2[i], result3[i]):\n",
    "        result.append('virginica')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "d1dd2c8c",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "accuracy:  0.8666666666666667\n"
     ]
    }
   ],
   "source": [
    "from sklearn.metrics import accuracy_score\n",
    "print(\"accuracy: \",accuracy_score(y_test, result))"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.10.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}