I have an issue with my code:
PCA runs fine, but when I combine it with grid search I get this error:
This is my code:
pca = PCA()
param_grid = {'n_components': range(1,50)}
GridSearchCV(pca, param_grid, cv=5 ,scoring='f1',
return_train_score=True, n_jobs=-1,
error_score="raise").fit(prepared_data_train, y_train)
CodePudding user response：
checkout this:
remove scoring as F1 does not make sense for PCA by default.
Also, make sure to normalize your data before performing PCA; the reason is explained here.
pca = PCA()
scaler = StandardScaler()
pipe = Pipeline(steps=[("scaler", scaler), ("pca", pca)])
param_grid = {'pca__n_components': range(1,50)}
search = GridSearchCV(pipe, param_grid, cv=5 , return_train_score=True, n_jobs=-1, error_score="raise").fit(X_digits, y_digits)
print("Best parameter (CV score=%0.3f):" % search.best_score_)
print(search.best_params_)
here is a good example from documentation for further exploration.