-
Book Overview & Buying
-
Table Of Contents
The Kaggle Book - Second Edition
By :
The core algorithms for hyperparameter optimization, found in the scikit-learn package, are grid search and random search. Recently, the scikit-learn contributors have also added the halving algorithm to improve the performances of both grid search and random search strategies.
HalvingRandomSearchCV is still experimental as of scikit-learn version 1.5.2, hence the API might suddenly change without any previous deprecation warnings. To use it, you need to explicitly import enable_halving_search_cv: from sklearn.experimental import enable_halving_search_cv.
In this section, we will discuss all these basic techniques. By mastering them, you will obtain practical optimization tools for specific problems (for instance, support vector machines or SVMs are usually optimized by grid search) and become familiar with the basics of hyperparameter optimization.
To start with, it is crucial to figure out what the necessary ingredients are:
...