🏠

1

2

3

4

5

6

8

Tuning

Feature Engineering and Selection: A Practical Approach for Predictive Models - online book

This needs to be one of the crucial steps where you would be moving closer to your final solution. Major steps should include:

• Test ensemble methods like voting classifiers etc.
• Hyperparameter tuning using cross-validation
• Use automated tuning methods like random search or grid search to find out the best configuration for your best models
• Test the models with as much data as possible
• Once finalized, use the unseen test sample that we set aside, in the beginning, to check for overfitting or underfitting

*click the image to enlarge: The Center is the true value and the dots are predictions.