Member Attrition Prediction for Credit Union

Purpose of the Project The purpose of this project was to develop a predictive machine learning model to identify credit union members who were at risk of leaving. By analyzing historical member data, transaction behavior, and engagement patterns, the model enabled the credit union to take proactive retention actions, reduce churn, and improve long-term member relationships.

Collected and analyzed historical member data including transactions, product holdings, balances, loans, and service usage over a defined time period.

Built a predictive attrition model using machine learning techniques to classify members into Retained, At-Risk, and Churned segments.

Performed feature engineering to identify behavioral patterns impacting member churn.

Developed an interactive Power BI dashboard to monitor attrition trends and highlight key drivers influencing member exit.

Extracted, cleaned, and transformed data related to:

Created meaningful features such as:

Trained and evaluated machine learning models to predict member attrition.

Identified high-risk members using propensity scoring.

Optimized model performance using accuracy, precision, recall, and ROC metrics.

Designed a Power BI dashboard showcasing:

Enabled targeted retention campaigns for high-risk members.

Improved decision-making for relationship managers and marketing teams.

Helped reduce churn and increase member lifetime value.