The Capstone project is the challenge set by Microsoft after completed 10+ require courses for the Microsoft Professional Programme (MPP) in Data Science. The project considers how demographics, location, property type, lender, and other factors are related to whether mortgage application across the United States was accepted or denied. We trained a Catboost model on 500,000 mortgage loan applications and noticed that lender, applicant income, loan purpose, loan amount and county code have a significant effect on the mortgage loan approval.

The prediction results whose code file can be found here achieved a public score of 0.7330 out of the benchmark of 0.7350 and you can also read the executive report here.