chimaobi-okite/loan_default_prediction

Loan_Default_Prediction

This repository contains a jupyter notebook file of the submission I made to the Loan Default Prediction Challenge on Zindi see more here.

Task Description

SuperLender is a local digital lending company, which prides itself in its effective use of credit risk models to deliver profitable and high-impact loan alternative. Its assessment approach is based on two main risk drivers of loan default prediction:. 1) willingness to pay and 2) ability to pay. Since not all customers pay back, the company invests in experienced data scientist to build robust models to effectively predict the odds of repayment.

These two fundamental drivers need to be determined at the point of each application to allow the credit grantor to make a calculated decision based on repayment odds, which in turn determines if an applicant should get a loan, and if so - what the size, price and tenure of the offer will be.

There are two types of risk models in general: New business risk, which would be used to assess the risk of application(s) associated with the first loan that he/she applies. The second is a repeat or behaviour risk model, in which case the customer has been a client and applies for a repeat loan. In the latter case - we will have additional performance on how he/she repaid their prior loans, which we can incorporate into our risk model.

It is your job to predict if a loan was good or bad, i.e. accurately predict binary outcome variable, where Good is 1 and Bad is 0.

Ranking

This solution is currently ranked Ist on Zindi's leaderboard. You can see more about it on the notebook