Project Name

This project focuses on a leading online loan marketplace that specializes in facilitating personal loans, business loans, and financing for medical procedures. The platform provides borrowers with easy access to competitive interest rates through a streamlined online interface.

A significant challenge faced by lending companies is the risk associated with lending to applicants deemed 'risky.' This risk often results in credit loss, which occurs when a borrower defaults—refusing to repay the loan or absconding with the funds. In this context, customers labeled as 'charged-off' represent those who have defaulted on their loans, leading to substantial financial losses for lenders.

The primary objective of this case study is to identify these high-risk loan applicants using Exploratory Data Analysis (EDA). By accurately detecting potential defaulters, the marketplace can reduce the number of risky loans issued, ultimately minimizing credit loss and enhancing financial stability. Through this analysis, we aim to contribute valuable insights that can inform lending strategies and improve risk management practices within the company.

Table of Contents

• General Info
• Technologies Used
• Conclusions
• Acknowledgements

General Information

• Background: This project leverages data analytics to improve lending practices within the online loan marketplace.

• Business Problem: The project aims to address the issue of credit loss by identifying risky loan applicants, thereby reducing the number of defaults and financial losses.

• Dataset: The analysis is conducted using a dataset that includes loan amounts, borrower income, loan status, and various borrower attributes.

Conclusions

• Loan Amount Analysis: The loan amounts taken by "charged-off" customers are comparable to those of "fully paid" customers, indicating that loan amount alone is not a reliable predictor of default.

• Income Discrepancy: The average income of charged-off customers is significantly lower than that of fully paid customers. Most customers who default earn less than $50,000, and those earning above $75,000 have a notably lower incidence of default.

• Housing Situation: A higher proportion of customers who default on loans reside in rented accommodations compared to fully paid borrowers.

• Loan Grade Correlation: Customers with grades B, C, and D are more likely to default than those with grades A, E, F, and G.

• Source Verification Impact: Customers whose "source" is verified show a lower tendency to default.

• Grade G Default Rate: Despite having higher annual salaries, customers with grade "G" show a higher rate of loan defaults compared to other grades.

• Tenure Influence: Loan tenure (30/60 months) does not significantly affect the likelihood of default.

• Correlation Insights: Loan amount, funded amount, total payment, and installment show a positive correlation, while amounts and public records are negatively correlated. Other attributes appear to have minimal dependent impacts on each other.

Technologies Used

• Pandas - latest version
• Matplotlib - latest version
• Seaborn - latest version

Acknowledgements

• N/A

Contact

• Siddhardha Mudumba
• Siju Babu