README.md

DIABETES-FEATURE-IMPORTANCE

In this repository, I explored what was the most important metrics lead to the severity of diabetes. Using ensemble methods (Random Forest and Gradient Boost), the two important features were body mass index (BMI) and s5 (ltg, possibly log of serum triglycerides level).

Learning fact: decision trees (and hence ensemble methods) are almost unaffected by scaling as they are insensitive to variance