对sklearn的使用之数据集的拆分与训练详解(python3.6)

from python/335560.html">sklearn import datasets#从cross_validation导入会出现warning，说已弃用from sklearn.model_selection import train-test_splitfrom sklearn.grid_search import GridSearchCVfrom sklearn.metrics import classification_reportfrom sklearn.svm import SVCimport sklearn.exceptioins#导入鸢尾花数据集iris = datasets.load_iris()#将数据集拆分为训练集和测试集各一半#其中X为数据特征（花萼、花瓣的高度宽度），为150*4的矩阵#Y为鸢尾花种类（0, 1, 2三种），为150*1矩阵#如果使用标签二值化， 将0, 1, 2表示为100 010 001#使用y.label_binarize(y, classes[0, 1, 2]),变为150*3矩阵X_train, X_test, y_train, y_test = train_test_split(iris.data, iris.target, test_size=0.5, random_state=0)#set the parameters by cross_validationturn_parameters = [{'kernel' : ['rbf', 'gamma' : [1e-3, 1e - 4, 'C':[1,10,100,1000]}, {'kernel':['linear'], 'C':[1,10,100,1000]}]#clf分离器#使用网格搜索法调超参数#训练集做5折交叉验证clf = GridSearchCV(SVC(C=1), turned_parameters, cv=5, scoring='%s_weighted' % score)#用前一半train数据再做5折交叉验证#因为之前的train_test_split已经分割为2份了#fit-拟合clf.fit(X_train, y_train)#超参数print(clf.best_params_)#得分for params, mean_score, scores in clf.gird_scores_: print("%.3f (+/-%.0.03f) for %r" % (mean_score, scores.std()*1.96,params))#分类报告y_true, y_pred = y_test, clf.predict(X_test)print(classification_report(y_true, y_pred))