Want to get involved? We're always looking for ideas and content for Weekly Challenges.
SUBMIT YOUR IDEA
Given no variables has correlation p-value less than 5%, so I used all columns (except TO as too many null values). Then for the regression model, I've used regularized model to reduce columns numbers.
Data Preparation is key here. From experience I've learnt not just to solve this problem, but make the solution dynamic that the data can be used to solve other problems.