Predicting mortgage default risk - 4th

This repo contains our work for SFU Business Analytics Hackathon. Here we are trying to predict mortgage default risk and identify key customer demographics to target in upcoming advertising campaign.

Modelling

• We predict the default risk using an ensemble classifier. This classifier was built in under 3 hours (length of hackathon) avergaging the predicted class probabilities output by a neural net and random forest.
• We used k-means clustering to identify key customer demographics to target in upcoming advertising campaign.

Team members:

• Nikola Surjanovic
• Claudine Tu
• Zubia Mansoor

Supplemental Materials