Skills & Tools Used:
✔ Power BI, Pivot Tables (Excel) – Data Visualization
✔ R (C4.5 Algorithm) – Decision Tree Classification
✔ Confusion Matrix, Accuracy – Model Evaluation
Project Overview
This project is the result of a practical work report conducted at Adira Finance, utilizing real company data to analyze credit applications over a four-month period. The study focuses on:
Descriptive analysis and data visualization using Pivot Tables and Power BI to identify trends in credit applications.
Classification of approved credit applications using the C4.5 decision tree algorithm, implemented in R to develop predictive rules for loan approval.
Key Insights from the Data
1,597 loan applications were approved between October 2022 – January 2023.
New Motorcycles had the highest demand, with 932 applications.
December & January had the most rejections, each with 135 declined applications.
More than 53% of loans were for multi-purpose financing, mainly for individuals and businesses with steady income.
How We Predicted Loan Approvals
By training a C4.5 decision tree, we discovered clear patterns in loan approval decisions:
✔ IF Status Document = Lengkap → THEN Approval = Approve
✔ IF Status Document = Save Partial AND Last Level Approval ≥ 2 → THEN Approval = Approve
✔ IF Status Document = Save Partial AND Last Level Approval < 2 AND Pihak Ketiga = Non-Dealer → THEN Approval = Approve
❌ ELSE → Approval = Reject
Model Performance
Training Accuracy: 96.93% – Strong performance on known data.
Test Accuracy: 97.17% – High reliability for new applications.
A clear, rule-based system that helps automate and streamline credit approvals.
Impact & Next Steps
This model reduces manual loan reviews, making approvals faster and more consistent. Future improvements could include additional financial indicators and machine learning ensembles for even better accuracy.
