如何解决嵌套 CV 中的规范化 - Python
我正在尝试在嵌套交叉验证中标准化我的数据。但是，我不确定我会在哪里做这件事。它可能需要在 cv for 循环中的某个地方，对吗？这是我的代码目前的样子：
X_train,X_test,y_train,y_test = train_test_split(x_final,target_df)
scaler = MinMaxScaler(feature_range=(0,1)) # <--- is this right here?
X_train = scaler.fit_transform(X_train) # <--- or should it go into the for loop?
X_test = scaler.transform(X_test) # <---
cv_outer = KFold(n_splits = 5,random_state = 21,shuffle=True)
for train_idx,vali_idx in tqdm(cv_outer.split(X_train,y_train)): #tqdm - progress bar
train_data,vali_data = X_train.iloc[train_idx],X_train.iloc[vali_idx]
train_target,vali_target = y_train.iloc[train_idx],y_train.iloc[vali_idx]
classifier = SVR()
cv_inner = KFold(n_splits=3,random_state=21,shuffle=True)
parameters = {'C':[10,100,200],'gamma':[0.01,0.1]}
grid_search = gridsearchcv(classifier,parameters,cv=cv_inner).fit(train_data,train_target)
best = grid_search.best_estimator_
best_classifier = best.fit(train_data,train_target)
y_pred = best_classifier.predict(vali_data)#[:,1]
r2 = r2_score(vali_target,y_pred)
print(r2,grid_search.best_score_,grid_search.best_params_)
# Final model
final_classifier = best.fit(X_train,y_train)
y_pred = final_classifier.predict(X_test)
有什么建议吗？
谢谢:)
版权声明：本文内容由互联网用户自发贡献，该文观点与技术仅代表作者本人。本站仅提供信息存储空间服务，不拥有所有权，不承担相关法律责任。如发现本站有涉嫌侵权/违法违规的内容， 请发送邮件至 dio@foxmail.com 举报，一经查实，本站将立刻删除。