A Telecom collaborates with an MFI to provide micro-credit on mobile balances to be paid back in 5 days. The Consumer is believed to be delinquent if he deviates from the path of paying back the loaned amount within 5 days.

Delinquency is a condition that arises when an activity or situation does not occur at its scheduled (or expected) date i.e., it occurs later than expected.
Many donors, experts, and microfinance institutions (MFI) have become convinced that using mobile financial services (MFS) is more convenient and efficient, and less costly, than the traditional high-touch model for delivering microfinance services. MFS becomes especially useful when targeting the unbanked poor living in remote areas. The implementation of MFS, though, has been uneven with both significant challenges and successes.
Today, microfinance is widely accepted as a poverty-reduction tool, representing $70 billion in outstanding loans and a global outreach of 200 million clients.

ndustry: Telecom.
Services: Mobile financial, Telecom Services.


Creating a model and analyzing delinquency can predict the probability for each loan transaction whether the customer will repay the borrowed amount within 5 days from the loan insurance date or not. (Label ‘1’ & ’0’)

Providing feedback to help companies minimize losses from overdue payment users


Data cleaning and validation

Analyzing the factors influencing delinquency and understanding the relationships between them.

Comparing and understanding customer loan profiles.

Check for the correlation with the dependent variable ‘Label..

Using machine learning techniques to build a model.


We can observe a group of users who are in delinquency and provide alerts and advice to the company.

Consider the percentage of users who can repay the loan amount.

In order to decrease loss to the company, the company should start some marketing strategies like sms alerting and notifications and others on the people with all loan levels and especially on low & high level people notifying them to pay the loan back within five days of time.

We can build a reliable probability prediction model with an accuracy of 89%.