Are Grid Search CV & Randomized Search CV always/necessarily supposed to give more accurate results after hyperparameter tuning as compared to K-Fold Cross Validation?
$\begingroup$
$\endgroup$
2
-
2$\begingroup$ Grid Search CV & Randomized Search CV are hyperparameter tuning strategies that use K-Fold Cross Validation internally to evaluate each hyperparameter configuration. Is this the intended question: does hyperparameter tuning consistently provide better model accuracy than not having hyperparameter tuning? $\endgroup$– grovOct 7, 2021 at 4:07
-
$\begingroup$ Yes, this is what I meant. I first tried K Fold CV on my ML model, then Grid Search. My accuracy turned out to be slightly higher for K Fold though Grid & Radom Searches are expected to do better. So my doubt is, is it possible for K Fold to perform better than Grid Search? $\endgroup$– ApoorvaOct 7, 2021 at 5:34
Add a comment
|
1 Answer
$\begingroup$
$\endgroup$
From your comment above, " though Grid & Radom Searches are expected to do better." They are EXPECTED to perform better but it is not a given that in each and every case they will outperform K Fold CV. Sometimes K FoldCV can outperform Grid or Random SearchCV.
