Supply Chain · Power BI
FMCG Supply Chain Management Analysis:
  • Experienced inefficiencies in inventory management, demand forecasting, and supplier distribution across multiple warehouse locations.
  • Identified regional demand gaps, supplier contribution imbalance, lead time variations, and mismatches between inventory and reorder levels.
  • Developed an interactive Power BI dashboard to optimize inventory planning, improve demand forecasting, and support efficient supply chain decision-making.
Power BI Excel Power Query EDA
View Insights
Real Estate · Python
Gurgaon Real Estate Market Analysis:
  • Faced difficulty in identifying pricing patterns, premium locations, and investment opportunities due to unstructured and inconsistent property data.
  • Identified high-value localities, pricing impact of RERA approval and property status, builder-driven price differences, and weak correlation between area and price per sqft.
  • Performed end-to-end EDA using Pandas, Seaborn, and Matplotlib to clean data, analyze trends, and generate actionable insights for real estate decision-making.
Python Pandas Seaborn Matplotlib EDA
View Project
Forecasting · Python
E-Commerce Marketplace Demand Forecasting Study:
  • Faced challenges in understanding demand patterns, forecasting accuracy, and seasonal variability across multiple product categories in e-commerce sales data.
  • Identified high predictability in grocery demand (~80% accuracy), high volatility and forecasting error in electronics, and increased sales variability during festive periods.
  • Applied a 30-day moving average forecasting model and built visualizations to evaluate MAPE, uncover seasonal trends, and support data-driven inventory and supply planning.
Python Excel Matplotlib Seaborn EDA
View Report
IPL Prediction · Supervised ML
IPL Match Win Prediction System:
  • Faced challenges in predicting match outcomes using complex ball-by-ball data, requiring effective feature engineering and handling of real-time match scenarios.
  • Identified that chasing teams had higher win probability (55.1%), logistic regression achieved the best performance (AUC: 0.760), and feature engineering had greater impact than complex models.
  • Built an end-to-end ML pipeline with EDA, feature engineering, multiple model comparison, and a real-time prediction system to estimate match outcomes based on first innings performance and contextual factors.
Pandas NumPy Scikit-learn Matplotlib Seaborn
View Project
Customer Segmentation · BI & AI Dashboard
Clickstream Behavioral Analysis:
  • Faced challenges in understanding user behavior, identifying high-intent customers, and reducing revenue loss due to cart abandonment and churn risk across the platform.
  • Identified key segments like buyers, browsers, and researchers, along with critical insights such as 63.1% cart abandonment, high CLV churn exposure, and strong impact of session depth and timing on revenue.
  • Developed an interactive AI Powered dashboard integrated with Excel-based analysis and AI-driven insights to enable user segmentation, revenue optimization, churn prevention, and targeted marketing strategies.
Power BI Excel Power Query AI Dashboard EDA
View Dashboard
Healthcare · Deep Learning Model
Breast Cancer Detection Neural Network:
  • Faced challenges in accurately classifying tumors as benign or malignant using medical diagnostic data while avoiding overfitting and ensuring reliable model generalization.
  • Identified strong class separation with high predictive performance (ROC-AUC: 0.9934, Accuracy: 98.25%) and minimal false negatives, highlighting the effectiveness of proper regularization and validation techniques.
  • Built a deep neural network with batch normalization, dropout, L2 regularization, and stratified cross-validation to deliver a robust and reliable cancer detection system.
Python TensorFlow Scikit-learn Matplotlib
View Project
Prediction · Supervised ML
Used Car Price Prediction System:
  • Faced challenges in predicting used car prices from messy real-world data with skewed distributions and multiple influencing features like age, fuel type, and usage.
  • Identified that proper data cleaning, log transformation, and feature engineering significantly improved performance, with Linear Regression achieving the best results (CV R² ≈ 0.73).
  • Built an end-to-end ML pipeline with preprocessing, feature engineering, cross-validation, and model tuning to accurately predict car prices and support data-driven valuation.
Python Scikit-learn Pandas Matplotlib
View Project