Our client, a prominent financial institution providing a wide array of financial services including accounting, lending, insurance, finance, loans, and online banking in the United States, approached us with a critical challenge. They aimed to proactively identify members who were at risk of missing their upcoming loan payments and forecast the probability of delinquency one month in advance, enabling them to take preventive measures.
Overview
Approach
To tackle this complex issue, we leveraged a comprehensive approach that made use of the available data:
- Loan-related data: This encompassed details such as the type of loan and loan amounts.
- Historical payment/advance transactions: We delved into the members’ past transaction histories to gain insights.
- Demographic and Credit Bureau data: This valuable information helped us understand the financial background of the members.
We embarked on a data exploration journey to discern the key variables that influenced delinquency. Additionally, we created supplementary variables, including:
- Total advances taken in the last 6 months: Understanding recent financial activities.
- Loan period remaining (till maturity): Gaining insights into the remaining loan duration.
- Total late fees incurred in the last 12 months: Assessing the impact of late payments.
With these insights in hand, we built a robust classification model using the data from the past 12 months.
Impact
Our solution delivered significant business impact. We successfully identified 80% of delinquent members within the top 2 probability deciles, which translates to the top 20% of high probability targets. This predictive power allowed our client to proactively manage and address potential delinquencies, ultimately safeguarding their financial stability and customer relationships.
