This is a Data Mining project for the course of Computational Learning. The goal is to use existing data of a bank to identify good and bad clients, as well as to predict if a loan would be paid back.
To do so, we've followed the steps of the CRISP-DM methodology: Business Understanding, Data Understanding, Data Preparation, Modeling, Evaluation, and Deployment.
Several models were tested and compared. Furthermore, we've experimented a lot of different techniques of Data Understanding, Preparation, Modeling and Evaluation.
The best model and parameters yielded an AUC-ROC score of 95.1%
The work done is described in detail in the final report.
- Bruno Rosendo (up201906334@fe.up.pt)
- João Mesquita (up201906682@fe.up.pt)
- Rui Alves (up201905853@fe.up.pt)