# 划分训练集和测试集,33%数据是测试集
data_train,data_test,label_train,label_test=train_test_split(
data,labels,test_size=0.33,random_state=42)
# scikit-learn的Simplelmputer可以进行缺失值的替换
imp = Simplelmputer(missing_values=np.nan,strategy='mean')
pd.DataFrame(imp.fit_transform(x))