Data Science Portfolio
Explore production-grade machine learning projects with proven business impact. From fraud detection to medical imaging, each project demonstrates real-world value.
Showing 17 of 17 projects
Domain
Technology
Industry
Retail Sales Forecasting
LightGBM achieving 13.5% WAPE for 16-day forecasting across 54 stores with 56 engineered features
Credit Card Fraud Detection
Real-time fraud detection using XGBoost and SHAP achieving 97% ROC-AUC and $131K savings per 100K transactions
COVID-19 X-ray Classification
Deep learning for medical imaging with 92% ROC-AUC and Grad-CAM explainability for clinical decision support
E-Commerce Customer Churn Prediction
Cost-optimized machine learning with 98% ROC-AUC and 60% cost reduction through threshold optimization
Financial Sentiment Analysis
NLP platform with BERT-MPNet achieving 81% accuracy on 4,846 expert-annotated financial sentences
Market Basket Analytics
Association rule mining and customer segmentation with 105K+ rules generating 20-35% increase in average order value
Stock Market Intelligence Platform
LSTM deep learning and technical analysis platform with 98.33% prediction accuracy across 7,195 US stocks
Retail Vision Analytics
Computer vision for fashion retail with 92% accuracy achieving 90% reduction in manual data entry time
Job Change Prediction (HR Analytics)
Cost-optimized HR analytics with 79.7% ROC-AUC and 20% cost reduction through business-aligned ML
Enterprise NER Intelligence (CoNLL-2003)
Named Entity Recognition with BERT achieving 92% F1-score for financial, legal, and healthcare applications
S&P 500 Intelligent Forecasting
Time series forecasting and portfolio optimization with Prophet and Modern Portfolio Theory for S&P 500
Twitter Sentiment Analysis
NLP platform with 88.3% ROC-AUC analyzing 400K tweets for brand monitoring and market intelligence
House Price Prediction
CatBoost regression achieving 90.4% R² accuracy with 223 engineered features and SHAP explainability
Car Insurance Premium Analytics
Stacking ensemble achieving 99.78% R² accuracy with real-time premium prediction and explainable AI
Wine Clustering Analysis
GMM vs K-Means comparison achieving 0.898 ARI with automatic cluster selection and uncertainty quantification
Naive Bayes Spam Detection
MATLAB-based email classifier achieving 91.6% ROC-AUC with hyperparameter optimization and bootstrap validation
COCO Smart Analytics
Computer vision platform with Faster R-CNN detecting 80+ object categories for retail, smart cities, and security