Feature Importance Explained - Main Models

With video explanation | Data Series | Episode 14.1

Overview

1 - What is feature importance?
2 - Uses of feature importance
3 - Linear and Logistic Regression feature importance
4 - Decision Tree, Random Forest and Gradient boosted trees
5 - Neural networks feature importance

1- What is Feature Importance?

In short, feature importance gives a score or metric that indicates how important a feature is in predicting a target variable. When constructing a machine learning model it is useful to be able to interpret the model that we have built to understand which features contribute to our target variable prediction the most or least.

For example: let us say that we used the number of bedrooms, garden size, distance to the closest school and distance to the closest shop to predict house price. To understand which of our features (the number of bedrooms, garden size, distance to the closest school and distance to the closest shop) contributes the most to a house price we can construct a model and look at the feature importance of each and identify which is the most or least important feature in predicting our house…