Project Portfolio

Home loan default prediction analyzing 58.44M records across 7 datasets. XGBoost classifier with 143 engineered features achieves 83% accuracy, 52.8% recall, and 0.785 ROC AUC at threshold 0.60. Addresses class imbalance (8% default rate), memory optimization (68.5% reduction), multicollinearity, and saves 1200+ hours of manual review through intelligent risk assessment.

Python XGBoost Scikit-learn Jupyter

Employee performance prediction using XGBoost multiclass classification (92.5% accuracy, 93.3% CV F1-score) with SHAP interpretability. Analyzes 1,200 employee records across 28 features, identifies top 3 performance drivers, and provides HR recommendations. Full pipeline includes EDA, feature engineering, model comparison, and deployment-ready inference.

Python XGBoost SHAP Scikit-learn

B2B sales lead quality prediction using XGBoost classifier. Achieves 81.06% ROC AUC and 84.74% recall on 7,420 IT sales leads. Handles class imbalance, high-cardinality categoricals, and missing data through frequency encoding and threshold optimization. Includes statistical analysis, cross-validation, feature importance, and actionable business insights.

Python XGBoost Scikit-learn Jupyter

Binary classification model for Portuguese bank term deposit prediction using LightGBM. UCI ML dataset (41,188 records). Handles class imbalance (7.87:1), multicollinearity (VIF>26k), and data leakage. Test: 87.8% accuracy, 57.3% recall, 81.1% ROC AUC with comprehensive cross-validation optimization for banking campaign targeting.

Python LightGBM Scikit-learn Jupyter

Daily bike rental demand prediction using Ridge Regression on Capital Bikeshare data (2011-2012). Addresses multicollinearity, zero-inflated features, and non-normal distributions. Test R²=0.832, CV R²=0.815±0.032. Includes statistical analysis, VIF removal, and comparison of 10 regression algorithms for accurate demand forecasting.

Python Ridge Regression Scikit-learn Jupyter

Machine learning regression model predicting 1985 automobile prices. Lasso model achieves 91.7% R² with superior generalization over XGBoost. Handles extreme multicollinearity (VIF 16,676→8.36), data leakage detection, and outlier treatment through PCA and domain-driven feature engineering with comparison of 10 regression algorithms.

Python Lasso PCA Scikit-learn

Desktop application management tool with intelligent window orchestration. Manage and arrange multiple applications across monitors with customizable grid and side-by-side layouts. Windows 10/11. Enhances productivity through smart workspace organization, automated application control, and multi-monitor support.

Electron Python Windows Automation

Automated CSV data analysis with statistical profiling and visualization. Features interactive CLI, automatic delimiter detection, memory-efficient processing, statistical profiling, and visualization generation. Supports Python 3.8-3.13. Published on PyPI v2.0.0 with comprehensive documentation.

Python 3.8+ CLI Pandas Rich

Static portfolio website using vanilla JavaScript ES6 modules and Bootstrap 5. Component-based architecture with dynamic nav/footer loading, modular CSS organization, Intersection Observer for animations, ESLint, Stylelint, Prettier for code quality, Lighthouse CI for performance monitoring, and optimized GitHub Pages deployment.

JavaScript Bootstrap 5 GitHub Pages ESLint