sklearn中的GridSearchCV如何选择交叉验证集？

Question

svc = GridSearchCV(SVC(), param_grid,scoring='f1_macro', verbose=1000)
svc.fit(X_train, y_train)
predictions = svc.predict(X_test)

I run this code many times but the results are the same. Does GridSearchCV choose the same cross-validation sets(not randomly choose)?

Answer 1

GridSearchCVKFold默认情况下使用交叉验证器，并且默认情况下KFold不打乱数据。要启用改组，您必须执行

from sklearn.model_selection import KFold

cross_validator = KFold(shuffle=True)
svc = GridSearchCV(SVC(), param_grid, scoring='f1_macro', 
                   verbose=1000, cv=cross_validator)
svc.fit(X_train, y_train)
predictions = svc.predict(X_test)