我已经对我的数据集进行了递归特征消除，现在试图基于RFE返回的特征进行预测，但一直遇到这个错误：

ValueError: X has 31 features per sample; expecting 9

这是我编写的获取最佳特征并根据返回的特征转换数据的代码

no_list = np.arange(1,len(list(dat)))acc_score = 0n_features = 0score_list = []for x in range(len(no_list)):    X_train,X_test,y_train,y_test = train_test_split(X,y,test_size = 0.2,random_state=100)    log_reg = LogisticRegression()    rfe = RFE(log_reg,no_list[x])    X_train_rfe = rfe.fit_transform(X_train,y_train)    X_test_rfe = rfe.transform(X_test)    log_reg.fit(X_train_rfe,y_train)    score = log_reg.score(X_test_rfe,y_test)    score_list.append(score)    if(score > acc_score):        acc_score = score        n_features = no_list[x]rfe = RFE(log_reg,n_features)rfe.fit_transform(X_train,y_train)predictions = rfe.predict(X_test)

回答：

在将X_test发送到预测函数之前，先对其进行转换。你的rfe使用了log_reg模型，该模型只接受n_features个特征。

rfe = RFE(log_reg,n_features)rfe.fit_transform(X_train,y_train)X_test = rfe.transform(X_test)predictions = rfe.predict(X_test)