Tobigs_Week7_SVM_Assignment.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "9ad88f3e",
   "metadata": {},
   "source": [
    "# Multiclass SVM 구현"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "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": 16,
   "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": 17,
   "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": 18,
   "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": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "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": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "11719030",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "     setosa  versicolor  virginica\n",
      "110   False       False       True\n",
      "69    False        True      False\n",
      "148   False       False       True\n",
      "39     True       False      False\n",
      "53    False        True      False\n",
      "..      ...         ...        ...\n",
      "64    False        True      False\n",
      "91    False        True      False\n",
      "81    False        True      False\n",
      "51    False        True      False\n",
      "0      True       False      False\n",
      "\n",
      "[120 rows x 3 columns]\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "(120, 3)"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# One-hot encoding\n",
    "y_train = pd.get_dummies(y_train)\n",
    "\n",
    "print(y_train)\n",
    "y_train.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "3a5cae81",
   "metadata": {},
   "outputs": [],
   "source": [
    "# OVR 사용 \n",
    "\n",
    "# class 0 인 경우  vs 나머지\n",
    "svm_1 = SVC(kernel ='rbf', C = 5, gamma = 5)\n",
    "# class 1 인 경우  vs 나머지\n",
    "svm_2 = SVC(kernel ='rbf', C = 5, gamma = 5)\n",
    "# class 2 인 경우  vs 나머지\n",
    "svm_3 = SVC(kernel ='rbf', C = 5, gamma = 5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "9772aef9",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[False False False False False False  True False  True False False  True\n",
      "  True False False False  True False False False  True False False False\n",
      " False  True  True False False False]\n",
      "[ True  True  True False False False False False False  True False False\n",
      " False False  True  True False False False False False False  True  True\n",
      " False False False False False  True]\n",
      "[False False False False  True  True False  True False False  True False\n",
      " False  True False False False False False  True False  True False False\n",
      "  True False False  True  True False]\n"
     ]
    }
   ],
   "source": [
    "# 학습\n",
    "svm_1.fit(X_train, y_train.iloc[:,0])\n",
    "svm_2.fit(X_train, y_train.iloc[:,1])\n",
    "svm_3.fit(X_train, y_train.iloc[:,2])\n",
    "\n",
    "print(svm_1.predict(X_test))\n",
    "print(svm_2.predict(X_test))\n",
    "print(svm_3.predict(X_test))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "id": "8cb73d5b",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The Score is same at index[3].\n",
      "The Score is same at index[17].\n",
      "The Score is same at index[18].\n"
     ]
    }
   ],
   "source": [
    "# 부호가 모든 같은 경우가 있는가? -> 모두 동점인 경우가 생긴다.\n",
    "for i in range(len(X_test)):\n",
    "    # 값의 부호를 이용해 모두가 동점인 경우가 있는지 출력하고자 한다.\n",
    "    if (np.sign(svm_1.decision_function(X_test)[i]) == np.sign(svm_2.decision_function(X_test)[i])) and (np.sign(svm_2.decision_function(X_test)[i]) == np.sign(svm_3.decision_function(X_test)[i])):\n",
    "        print(f\"The Score is same at index[{i}].\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "id": "08316007",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "prediction : \n",
      " ['versicolor', 'versicolor', 'versicolor', 'virginica', 'virginica', 'virginica', 'setosa', 'virginica', 'setosa', 'versicolor', 'virginica', 'setosa', 'setosa', 'virginica', 'versicolor', 'versicolor', 'setosa', 'versicolor', 'virginica', 'virginica', 'setosa', 'virginica', 'versicolor', 'versicolor', 'virginica', 'setosa', 'setosa', 'virginica', 'virginica', 'versicolor']\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "0.8666666666666667"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# decision function을 이용해서 동점 처리\n",
    "# decision function : SVM 모델의 결정 함수 값 반환 (결과 : 클래스 경계로부터의 거리)\n",
    "\n",
    "def one_rest_svm_predict(models, data, labels): # 예측 수행 함수\n",
    "    distance = None # 초기화 진행\n",
    "    \n",
    "    for model in models: # 주어진 모든 svm 모델에 대해 반복\n",
    "        if distance is None: # 초기화 안 되어 있을 경우\n",
    "            distance = model.decision_function(data) # decision_function으로 초기화\n",
    "        else:\n",
    "            distance = np.vstack((distance, model.decision_function(data))) # 모델이 비어있지 않으면 배열 stack (수직방향으로)\n",
    "            # 중요! 쌓이는 데이터의 차원(데이터의 열 수는 동일해야 한다.)\n",
    "            \n",
    "    # 반복문 수행 후 distacne 의 row에는 모델의 수 , columns 에는 len(data) 가 저장된다.\n",
    "    # 계산의 편의를 위해 Transpose // 행이 데이터 샘플을 나타내도록 변경\n",
    "    distance = distance.T\n",
    "    \n",
    "    result = [] # 예측 결과값에 대한 distance를 저장하는 배열 생성\n",
    "    \n",
    "    for pred in distance:\n",
    "        result.append(labels[pred.argmax()]) \n",
    "        # 각 예측들에 대해서 가장 큰 값을 가지는 결과값(argmax) 에 해당하는 꽃을 저장\n",
    "\n",
    "    \n",
    "    print(f\"prediction : \\n {result}\") # 예측 결과 출력\n",
    "    return result\n",
    "\n",
    "models = [svm_1, svm_2, svm_3]\n",
    "labels = [\"setosa\", \"versicolor\", \"virginica\"]\n",
    "\n",
    "prediction = one_rest_svm_predict(models, X_test, labels) # 함수 호출을 통한 예측 수행\n",
    "\n",
    "accuracy_score(y_test,prediction) # 정확도 계산"
   ]
  }
 ],
 "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": "3.11.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}