Customer Credit Behaviour Analysis

Project Goal & Purpose

The main goal of this project was to help financial institutions predict loan default risks and understand customer behavior using real-world data.

By analyzing clients’ financial, employment, housing, and payment history, we aimed to uncover patterns that explain why some customers fail to repay on time while others stay consistent.

Our objective wasn’t just technical — it was about turning raw data into business value.

We wanted to support smarter lending decisions, reduce financial losses, and promote fair, data-driven credit approval processes.

What We Did:

Data Cleaning & Preparation (Python):

We handled missing values, standardized formats, and created new features such as Age, Years of Employment, and Active Loan Flags.

Data Modeling (Power BI):

Built a relational data model connecting 7 cleaned datasets to ensure accurate cross-table analysis.

Visualization & Insights:

Developed interactive dashboards showing demographic, financial, and behavioral factors that affect repayment performance.

Key Insights:

The overall default rate is ≈ 8%, showing most clients repay on time.

Behavioural indicators (late payments, job type, marital status) predict risk better than income or loan amount.

Married and older clients are more financially disciplined.

Housing stability (owning vs renting) strongly correlates with repayment reliability.

Past late payments are the strongest early warning sign for future defaults.

Recommendations:

Implement early-warning systems for customers showing delay trends.

Segment clients by repayment behaviour and income stability.

Offer flexible repayment terms for unstable income groups.

Use machine learning to predict default risk before it happens.

A massive thank you to our supervisor Eng. Huda Hemdan for her continuous support, valuable guidance, and encouragement throughout the project