Feature Selection

Topics

This week’s assignments will guide you through the following topics:

• How to extract feature importance values from your trained model
• How to use that information to perform feature selection

Reading

Please read the following:

• Supplemental Reading Feature Importance and Feature Selection With XGBoost in Python

Replication task

• Extract and plot feature importance values for your model
• Iterate through different importance threshold values to select a less complex model that maximizes accuracy

Tasks

Complete the following tasks:

• Read the article on feature importance and feature selection
• Apply the techniques in the article on your own model
• Complete the weekly questions below
• Work on project proposal and elevator pitch assignments

Weekly Questions

Answer the following questions

• What does feature importance tell us? Why is that useful to know?
• What’s the difference between ‘gain’ and ‘weight’ when it comes to feature importance for tree-based models?
• Why can reducing the total number of features in our model be helpful or preferable?