Risk Profiling of Customers for Installment Loans in Lending Business

About Client

The client is Canada’s largest consumer finance retailer operating hundreds of stores across the country offering a wide range of financial services for their customers.

The Background

Credit risk analysis is undertaken by almost all financial institutions. It is one of the most crucial elements to ensure stability in everyday functioning. Unsecured short-term loans are among the most popular credit products in the world. Not to mention the fierce competition in the finance space, managing any lending activity requires powerful analytics solutions that can provide adequate information about customers and business performance.

The Challenge

As a part of their ongoing efforts to increase their revenue and market presence, the client was gearing up to launch their new product, installment loans. To succeed in their endeavor, the client needed an analytics solution that could predict defaulters and non-defaulters, predict insolvency in time and at the same time lower their NPA levels.

Our Solution

  • In order to meet the objective, the situation called for exploratory data analytics and thin file analysis for new customers and identifying necessary variables from the ocean of customer data procured from Credit Unions, forms filled by the customers, and more.
  • As a result, over 500 variables were identified. For high accuracy, Machine Learning algorithms were deployed to filter the key variables from the dataset.
  • Multiple Machine Learning models were built to predict delinquency and insolvency.
  • Feature extraction mechanism was used to select important variables based on weightage.
  • Various Data Visualization techniques were tested and reports generated for discovering trends in customer behavior, product performance, store performance and to identify customer delinquency patterns.
  • Delinquency trends were identified based on locations using Power BI, SQL server, and MS-Excel
  • High loan balances were marked as threat to identify fraud or high-risk customers
  • Competitors data were obtained from Credit Bureaus for reverse engineering and extracting more information on customers
  • Machine Learning algorithms were deployed for segmenting customers based on their scores
  • As a final step, impact analysis of a product against the performance of other products for ensuring optimal revenue generation

The client recorded

increase in Loan book within a quarter as their approval rates improved.

They were able to identify the best and worst performing loan products.

They were able to predict defaulters, non-defaulters, and insolvency which significantly reduced their NPA.

Recent Case Studies

Get in Touch