python 实现SVM算法

from sklearn import svm, metrics
import pandas as pd
from sklearn.metrics import classification_report
from sklearn.model_selection import train_test_split
import numpy as np
import matplotlib.pylab as plt
def do_metrics(y_test,y_pred):

    print ("metrics.accuracy_score:")
    print (metrics.accuracy_score(y_test, y_pred))
    print ("metrics.confusion_matrix:")
    print (metrics.confusion_matrix(y_test, y_pred))
    print ("metrics.precision_score:")
    print (metrics.precision_score(y_test, y_pred))
    print ("metrics.recall_score:")
    print (metrics.recall_score(y_test, y_pred))
    print ("metrics.f1_score:")

    print (metrics.f1_score(y_test,y_pred))
    print("classification_report:")
    print(classification_report(y_test, y_pred))
    plot_auc(y_test,y_pred)
    # print("--->",type(y_test))
    # print("--->",type(y_pred))
    # print("metrics.auc:", metrics.auc(np.array(y_test), np.array(y_pred)))

#auc计算并生成图形
def plot_auc(y_test,y_pred):
    print("auc:")
    fpr, tpr, thread = metrics.roc_curve(np.array(y_test), np.array(y_pred))
    x=metrics.auc(fpr, tpr)
    print(x)
    plt.title("ROC curve of %s (AUC = %.4f)" % ('lightgbm', x))
    plt.xlabel("False Positive Rate")
    plt.ylabel("True Positive Rate")
    plt.plot(fpr,tpr)  # use pylab to plot x and y
    plt.show()  # show the plot on the screen
def svmModel():
    textFile=open(r'5-22-train.csv','rb')
    data = pd.read_csv(textFile,header=None,prefix='x', error_bad_lines=False,low_memory=False)
    data=data.dropna()
    # 打乱数据的分布
    data.sample(frac=1).reset_index(drop=True)
    data.rename(
        columns={
            'x0': 'userId', 'x1': 'vreN', 'x2': 'lireN', 'x3': 'repN','x4': 'ot', 'x5': 'appV', 'x6': 'age', 'x7': 'bage',
            'x8': 'keyword1', 'x9': 'keyword2', 'x10': 'tag1', 'x11': 'tag2', 'x12': 'tag3', 'x13': 'ks1', 'x14': 'ks2',
            'x15': 'historySubT','x16':'ttp1','x17':'ttp2','x18':'ttp3','x19':'ktw1','x20':'ktw2','x21':'ktw3','x22': 'apn',
            'x23': 'city_id','x24': 'mode', 'x25': 'phone_price', 'x26': 'city_level', 'x27': 'itemId', 'x28': 'itemType',
            'x29': 'label','x30': 'texL', 'x31': 'titL', 'x32': 'twN', 'x33': 'subT', 'x34': 'sour',
             'x35': 'imgs_count', 'x36': 'categordId','x37': 'status', 'x38': 'author_type'
        }, inplace=True
    )

    #删除"-"字符
    data=data.drop(data[data["city_id"] == "-"]["city_id"].index)
    data["city_id"]=data["city_id"].apply(lambda x:int(x))
    #训练集的标签
    y_train =data.label
        #所有的label减1
        #label==0表示曝光 label==1表示点击
    y_train=y_train[:]-1
    x_train = data.drop(['label'], axis=1).drop(labels="itemId", axis=1).drop(labels="userId", axis=1) \
            .drop(labels="itemType", axis=1)
    x_train=np.array(x_train)
    y_train=np.array(y_train)
    x_train, valid_train, y_train, valid_test = train_test_split(x_train, y_train, test_size=0.9,
                                                         random_state=1)  # 分训练集和测试集
    textFile = open(r'5-22-test.csv', 'rb')
    data1 = pd.read_csv(textFile, header=None, prefix='x', low_memory=False)
    # 打乱数据的分布
    data1=data1.dropna()
    data1.sample(frac=1).reset_index(drop=True)
    data1.rename(
        columns={
            'x0': 'userId', 'x1': 'vreN', 'x2': 'lireN', 'x3': 'repN', 'x4': 'ot', 'x5': 'appV', 'x6': 'age',
            'x7': 'bage',
            'x8': 'keyword1', 'x9': 'keyword2', 'x10': 'tag1', 'x11': 'tag2', 'x12': 'tag3', 'x13': 'ks1', 'x14': 'ks2',
            'x15': 'historySubT', 'x16': 'ttp1', 'x17': 'ttp2', 'x18': 'ttp3', 'x19': 'ktw1', 'x20': 'ktw2',
            'x21': 'ktw3', 'x22': 'apn',
            'x23': 'city_id', 'x24': 'mode', 'x25': 'phone_price', 'x26': 'city_level', 'x27': 'itemId',
            'x28': 'itemType',
            'x29': 'label', 'x30': 'texL', 'x31': 'titL', 'x32': 'twN', 'x33': 'subT', 'x34': 'sour',
            'x35': 'imgs_count', 'x36': 'categordId', 'x37': 'status', 'x38': 'author_type'
        }, inplace=True
    )

    # 删除"-"字
    data1 = data1.drop(data1[data1["city_id"] == "-"]["city_id"].index)
    data1["city_id"] = data1["city_id"].apply(lambda x: int(x))
    # 训练集的标签
    # 原始label
    y_test = data1.label
    y_test1 = y_test[:] - 1
    y_test1=np.array(y_test1)
    # 测试集数据
    x_test = data1.drop(labels="label", axis=1).drop(labels="itemId", axis=1).drop(labels="userId", axis=1) \
        .drop(labels="itemType", axis=1)

    x_test=np.array(x_test)
    model=svm.SVC(C=1,kernel='rbf',gamma=20)
    model.fit(x_train,y_train)
    model.score(x_train,y_train)
    ypred= model.predict(x_test)
    do_metrics(y_test1,ypred)
if __name__=="__main__":
    svmModel()