Practical AI Notes - for own use

Supervised Machine Learning Algorithms - Decision Tree

1. Decision Tree is a white box technique where the resulting model is interpretable.
2. Scaling is not necessary but categorical and missing data must be handled in data preprocessing stage.
3. Nonetheless, the default hyperparameter value, will lead to fully grown tree which will overfit.
4. Pre-pruning can be applied to limit the growth of the tree by setting the following hyperparameters:
• max_depth - Maximum depth of the tree 
• max_leaf nodes - Maximum of leaves of the tree.