PROJECT 02 · DATA ANALYSIS · SQL · EXCEL · POWER BI

Credit Risk Analysis — Loan Default Prediction

Analyzed 32,581 real-world loan records to identify the key factors that predict whether a borrower will default. Full analyst workflow: data cleaning and validation in Excel, 7 SQL queries for risk segmentation and borrower profiling, and an interactive Power BI dashboard with DAX measures, KPI cards, and traffic-light color scheme.

Complete SQL Excel Power BI DAX SQLite Kaggle Dataset

POWER BI DASHBOARD

Project Files

⌨

GitHub Repository Source files · README

→

🗄

SQL Queries 7 analytical queries

📊

Clean Dataset Excel · 32,581 records

⚡

Power BI Dashboard .pbix · Interactive

Key Findings

Finding 01

Loan Grade is the Strongest Default Predictor

Grade G loans default at 98.44% — nearly guaranteed non-payment. Grades A and B show default rates below 17%, confirming that credit grading accurately reflects risk. This validates the grading system as the primary filter for loan approval decisions.

Finding 02

Home Ownership Signals Repayment Behavior

Renters default at 31% vs 7% for homeowners — a 4x risk difference. This suggests that housing stability is a strong proxy for financial stability overall, and should be weighted heavily in lending decisions alongside income.

Finding 03

Income Gap Between Defaulted and Good Standing Borrowers

Borrowers in default earn 31% less on average than those in good standing ($49K vs $70K annually). Debt Consolidation loans show the highest default volume (1,490 loans at 28.59%), suggesting borrowers are already in financial distress at the time of application.

Analyst Workflow

Data Acquisition

Downloaded dataset from Kaggle — 32,581 loan records, 12 variables including borrower demographics, loan characteristics, and repayment status.

Data Cleaning in Excel

Verified 32,581 unique records, identified and removed 1 duplicate using Conditional Formatting, validated data types, confirmed no nulls in key columns, removed currency symbols for clean CSV export.

SQL Analysis

Wrote 7 analytical queries covering: portfolio split by status, risk tier classification by loan-to-income ratio, default rate by loan intent, borrower profile comparison, default rate by home ownership, loan grade risk classification, and high-income unexpected defaults.

Power BI Dashboard

Built two-page interactive dashboard: Portfolio Overview (KPI cards, default rate by grade, by home ownership, loan intent donut) and Key Insights (top 3 grades/intents by default rate, income comparison, risk tier distribution). Used DAX measures for dynamic calculations.

Business Recommendations

Documented 4 actionable recommendations: restrict Grade E–G approvals, prioritize homeowners in credit decisions, flag Debt Consolidation for enhanced review, and enforce income verification as primary risk indicator.