0
$\begingroup$

Currently I have a dataset with 4712 records focused on binary classification.

I have divided my data into train(70%) and test (30%) split.

I run a gridsearchcv on my train data and choose the best parameters to fit the model as shown below

param_grid = {'C': np.logspace(-3,3,7),'penalty':['l1','l2'],'max_iter':[100,200,300],'class_weight':['balanced'],'solver':['newton-cg','lbfgs','liblinear','sag','saga'] }
logreg=LogisticRegression(random_state=41)
logreg_cv=GridSearchCV(logreg,param_grid,cv=10,scoring='f1')
logreg_cv.fit(X_train_std,y_train)
y_pred = logreg_cv.predict(X_test_std)
cm = confusion_matrix(y_test, y_pred)

The above code works fine.

However if I include the bootstrap:[True] in the parameter list, I get an error message as shown below

ValueError: Invalid parameter bootstrap for estimator LogisticRegression(C=0.01, class_weight=None, dual=False, fit_intercept=True,

But I came to know about something called bootstrap validation through 6th point in first answer of this post.

Can someone help me understand how different is it from CV that I am doing above? Or when I did cross validation above, does it mean I have already done bootstrap validation?

Or does Bootstrap validation involves dividing the test data (30%) into multiple chunks?

Can someone help me with as to how it works?

$\endgroup$