The Magic of Grid Search Cross Validation

Introduction

GridSearchCV is one of the most popular hyperparameter tuning libraries in the world of data science. As the name suggests , it is used to tune the parameters for the models inorder to get more accurate predictions out of that model. This library can be used for any models even if that particular model is not part of the scikit-learn libraries (For example: Xg Boost model)

What are hyper-parameters??

Before we deep dive into the library itself , let’s look at what are hyper-parameters. Hyper-parameters are part of the default algorithm which we can change according to different datasets.

Why are hyper-parameters important for modelling purposes??

How does GridSearchCV comes into the picture??

At the end of the day we are only human beings. It is impossible to find the best match by trying out each hyper-parameters and looking for the best score. Gridsearch helps with hassle of trying out each parameters.

Like in this image , We can enter our own ‘range’ of hyper-parameters for the GridSearchCV, which inturn finds the best hyper-parameters out of the ones we entered.

How to use GridSearchCV??

After defining those hyper-parameters in a variable:

If we closely look at this method of doing gridsearch , before introducing the gridsearch itself we first defined the hyper-parameters that we are going to input and the model we are trying to find the best hyper-parameters through GridSearchCV. This is the best way to do GridSearchCV

How does GridSearchCV works??

If you want to know more about cross validation, you can refer to the following medium article:

What are the hyper-parameters behind GridSearchCV??

As you saw earlier in the image, there are several hyper-parameters which are present in the GridSearchCV itself. Let’s jump into some of those in detail:

Advantages and Disadavntages of GridSearchCV

Advantages

Disadvantage

Conclusion

At the end of the day , it is upto each individual’s choice to opt for GridSearchCv or not. But there is no argument that GridSearchCV is one of the most used and important part of the data science world. For a layman , GridSearch is actually magic!!

Alternatives

If you are not content with GridsearchCV there are alternatives for this. Those alternatives are:

Thses are the main alternatives available within the scikit-learn world. Due to the open-sourced nature of python, there might be other alternatives for GridSearchCV outside the scikit-learn.

References for the research

Happy Reading!!!

Related posts:

ChatGTP offers us a completely new way to explore classical music. Not only can you can ask it to analyze compostions but you can start a dialog with the composer him or herself

The young and brilliant rapper, Logic wrote a song called “Anziety” talking about his struggles coming to terms with an anxiety attack that he had. The physical effects can be harrowing to deal with…