Contributed By – Amresh Kumar PGPDSBA Alumnus
I work as a Business Analyst for a fin-tech company which is a small finance bank primarily providing services in Retail loans to consumers and MSME (Micro small-medium enterprise) loans to small business. Most of the loan applications are reviewed manually and approval status is generated basis profitability and the loan’s default risk. Since the organization is small, most of the operations are manual but considering the organization’s growth and increasing loan book size, there is a need to automate the loan review process.
I worked on multiple loan data be it home, personal, vehicle and educational loans that were approved by the bank in past one year. Loan information attributes such as loan amount, term, interest rate, loan type along with applicant’s information like income, existing loans, past payment history and CIBIL score were analyzed to identify the relationship with default. A Machine Learning model was developed on this data to predict default and generate a credit score. The score was able to capture most of the defaults with high accuracy.
I tested the model on recent loans applications and it reduced our manual efforts by 80% declining the riskiest applications automatically.
I am also manually reviewing a few of the applications and tracking them to be sure of model performance. This model will also reduce the scope of human error and do root cause analysis of why certain loans are approved or declined.
If you wish to learn more such concepts and work on interesting capstone projects, take up the PGP Data Science and Business Analytics Course by Great Learning and upskill today.
