- Developed and trained model with team to develop a predictive model to Bike Rental Counts
- Performed EDA using Tidyverse and mitigated multicollinearity by removing redundant features with VIF > 10
- Utilized Exhaustive search to identify the best features using model metrics, such as Adj R2, Mallow's Cp and BIC, reducing model from 20 to 13 variables without decreasing model performance.