Decision Tree Point's to keep with implementations

You will go through important points of Decision Tree

You will go through my notebook to see how to apply Decision Tree end2end.

Lets dive in:

How to find Purity of Nodes?

• Entropy - use when features are less || more entropy more uncertainty || ranges (0 to 1)

• Gini Impurity - use when features are more || by default Decision Tree classifier uses gini impurity || ranges (0 to 0.5)

Gini Impurity for Binary Class

How to Select Feature to Split on?

• Information Gain helps to select feature to split on
• more information we gain the better feature
• use that feature as root node

Usually Decision Tree's tends to Overfit

How to Prevent Overfitting in Decision Tree?

1. Post Pruning
2. Pre Pruning

Hyperparameters to Train:

• max_depth
• max features
• n_estimators
• min_samples_split
• min_samples_leaf

Above given are quick points to recap how tree works, now if you can't relate much or maybe you want to see how to perform all of above points practically with Hyperparameter Tuning visit this notebook:

Next will share some interview questions on Decision Tree.

????????????????????????????????????????????????

Happy Learning??