SVM.py

# -*- coding: utf-8 -*-
# @Time    : 2021/9/17 7:49 下午
# @Author  : Yushuo Wang
# @FileName: SVM.py
# @Software: PyCharm
# @Blog    ：https://lesliewongcv.github.io/

import numpy as np
import os
from scipy.io import loadmat
from sklearn.model_selection import KFold
from sklearn.svm import SVC
import matplotlib.pyplot as plt

PATH = 'Data/'
df_train = loadmat(PATH + 'data_train.mat')['data_train']
df_test = loadmat(PATH + 'data_test.mat')['data_test']
df_label = loadmat(PATH + 'label_train.mat')['label_train'].squeeze()
K = 4
kf = KFold(n_splits=K, shuffle=False)
C_list = np.arange(0.5,4, 0.1)
GAMMA_list = np.arange(0.05,0.50, 0.05)
ACC = []
# for C in C_list:
for gamma in GAMMA_list:
    acc_train = 0
    acc_valid = 0
    for train_index, valid_index in kf.split(df_label):  # 4-fold

        model = SVC(C=1.5, kernel='rbf', gamma=gamma)
        model.fit(df_train[train_index], df_label[train_index])

        res_train = model.predict(df_train[train_index])
        res_valid = model.predict(df_train[valid_index])

        acc_train += float(sum(res_train == df_label[train_index]) / len(train_index))
        acc_valid += float(sum(res_valid == df_label[valid_index]) / len(valid_index))

        #print('ACC using SVM: ' + str(acc_train) + ' | ' + 'Valid: ' + str(acc_valid))
    ACC.append(acc_valid/K)
    # print('C: %.2f' % C)
    print('gamma: %.2f' % gamma)
    print('ACC using SVM: ' + str(acc_train / K) + ' | ' + 'Valid: ' + str(acc_valid / K))

plt.plot(C_list, ACC, c='red')
plt.ylabel('ACC')
plt.xlabel('C')
plt.title('Acc with different C on Validation set')
plt.show()