Runtee/DSN

Data Science Nigeria Bootcamp Qualification Competition # DSN Introduction This is a competition hosted by Data Science Nigeria as part of prerequisite for it's all-expenses-paid learning bootcamp. Hosted on Zindi zindi.africa' Competition Score: 0.844648358439733 Problem Statement Kowope Mart is a Nigerian-based retail company with a vision to provide quality goods, education and automobile services to its customers at affordable price and reduce if not eradicate charges on card payments and increase customer satisfaction with credit rewards that can be used within the Mall. To achieve this, the company has partnered with DSBank on co-branded credit card with additional functionality such that customers can request for loan, pay for goods even with zero-balance and then pay back within an agreed period of time. This innovative strategy has increased sales for the company. However, there has been recent cases of credit defaults and Kowope Mart will like to have a system that profiles customers who are worthy of the card with minimum if not zero risk of defaulting. Objective To predict customers who are likely to default or not. Model Used Catboost classifier Evauation Metric Area under the ROC Curve(AUC) Variable Definitions Application_ID: Unique Customer Application Identification number Risk Scoring features: form_field1: Customer Creditworthiness score based on historical data form_field2: A score that measures the number and riskiness of credit enquiries made by a borrower. Severity features: Severity is a function of amount, time since default, and number of defaults form_field3: Severity of default by the borrower on any loan(s). form_field4: Severity of default by the borrower on auto loan(s). form_field5: Severity of default by the borrower on education loan(s). Credit features form_field6: Minimum of credit available on all credit cards that is automatically renewed as debts are paid off on the customer’s cards (in NGN) form_field7: Maximum …