For the project I propose to analyze loan data from Mintos - the largest European-based retail lender. The project will allow me to develop better understanding of credit risk, and showcase my skills as a data scientist to potential employers. The project is relevant to almost all commercial companies, as every financial transaction carries some sort of credit risk. While other credit datasets have been studied, I could not find any reference to a thorough study of Mintos data.

The lending structure of Mintos is quite different from two other comparable platforms - Prosper and Lending Club mentioned on TDI blog. Almost all the loans come from 3rd party originators (in 37 countries and 5 continents) that guarantee return of the principal and interest in case of borrower default. In plain terms this means that primary credit risk lies with originators, or originator group, and not with individual borrowers. There has been only 1 originator default in Mintos history, so direct default modelling is not possible. Also, unlike Prosper and Lending Club Mintos has a thriving no fee secondary market, where vintage loans can be bought and sold.

These two structural factors make for a fascinating and complex dataset to analyze.