Apart from feature transformations mentioned in the previous section, each ML algorithm has its own hyperparameter space to be optimized. You can think of searching the best ML pipeline as going through your configuration space and trying out your options in a smart way to find the best performing ML pipeline.
Auto-sklearn is very helpful in achieving that goal and the example that you have seen in the introductory chapter showed you the ease of use of the library. This section will explain what's happening under the hood to make this implementation successful.
Auto-sklearn uses meta learning to select promising data/feature processors and ML algorithms based on properties of the given dataset. Please refer to the following links for the list of preprocessing methods, classifiers, and regressors:
- Classifiers (https://github.com/automl/auto-sklearn/tree/master...
