Back
Year
2025
Tech & Technique
Python, XGBoost, Scikit-learn, Pandas, NumPy, SMOTE, Flask, SQLite, Chart.js, HTML/CSS/JS, Data Analysis, Machine Learning
Description
Risk & Retention Intelligence Engine is a comprehensive machine learning platform that combines advanced banking customer churn prediction with real-time credit card fraud detection. This production-ready system demonstrates expertise in data analysis, ML engineering, and full-stack development through an executive dashboard with professional-grade analytics.
Key Features:
Technical Highlights:
Data Analysis Insights:
Key Features:
- Banking Churn Model: 91.6% accuracy with XGBoost + SMOTE balancing
- Fraud Detection: 96.0% accuracy with ensemble modeling and threshold optimization
- Executive Dashboard: Professional UI with real-time predictions and business insights
- Advanced Analytics: 30+ comprehensive visualizations covering demographics and risk patterns
- Production Architecture: Flask REST API with SQLite database and enterprise logging
- Business Intelligence: Geographic risk analysis, customer segmentation, and value tier analysis
Technical Highlights:
- Processed 10,000+ banking customers and 284,807 fraud transactions
- Engineered 64 features including demographic risk scores and behavioral patterns
- Implemented SMOTE oversampling with 95.4% precision for churn prediction
- Built interactive dashboard with Chart.js visualizations and real-time inference
- Deployed scalable architecture with comprehensive error handling and monitoring
- Created business impact analysis showing 87.4% customer retention efficiency
Data Analysis Insights:
- Geographic Risk: Germany (32.4%), Spain (16.7%), France (16.2%) churn rates
- Gender Analysis: Female customers show 25.1% vs male 16.5% churn rates
- Age Patterns: Peak churn in 45-49 age group (43.4%), lowest in 25-29 (7.1%)
- Product Impact: 3+ products correlate with 16.2% higher churn probability
- Behavioral Indicators: Inactive members show 26.9% vs active 14.3% churn rates
My Role
Data Analyst & ML Engineer
Led the complete development of this financial risk management platform:
Led the complete development of this financial risk management platform:
- Data Analysis: Performed comprehensive EDA on 294K+ records with 30+ visualizations
- Feature Engineering: Created 64 advanced features including risk scores and demographic analysis
- Model Development: Built and optimized XGBoost models achieving 91.6% and 96.0% accuracy
- Dashboard Development: Created executive-level dashboard with Flask backend and interactive frontend
- Business Intelligence: Delivered actionable insights on customer retention and fraud prevention
- Production Deployment: Implemented scalable architecture with monitoring and logging systems