Projects

Objective: Develop and deploy a machine learning model for predicting customer-level sales and profit generation, while optimizing inventory levels and ordering procedures based on data analysis. Tools and Technologies: •Python •Scikit-learn •Pandas •NumPy •SQL •Power BI Process: •Data Preprocessing: 1)Cleaned sales and inventory data by addressing missing values, duplicate features, and inconsistencies. 2)Performed exploratory data analysis (EDA) to understand data distribution, identify key relationships, and discover actionable insights. •Model Building: 1)Developed and trained a linear regression model using scikit-learn to predict customer-level sales and profit. 2)Evaluated model performance using metrics like Mean Absolute Error (MAE), Mean Squared Error (MSE), Root Mean Squared Error (RMSE), and R-squared. •Inventory Optimization: 1)Analyzed inventory data to identify reorder points and safety stock levels, ensuring efficient stock management and minimizing stockouts. 2)Developed algorithms to optimize ordering procedures based on factors like lead time, average order quantity, and demand forecasts. •Implementation: 1)Saved the trained prediction model using pickle for future use. 2)Designed and implemented data pipelines for automated data ingestion, model training, and prediction generation. 3)Developed dashboards and visualizations in Power BI for communicating insights to stakeholders. Results: •Achieved a Mean Absolute Error (MAE) of 236.279688, Mean Squared Error (MSE) of 144248.5382, and Root Mean Squared Error (RMSE) of 379.800656 for the sales prediction model. •R-squared value of 0.9463 indicates strong correlation between predicted and actual sales. •Identified key insights from the EDA, such as: 1)Revenue and profit decreasing despite increase in orders due to reduced basket quantity. 2)Over 50% of sales coming from Latin America and Pacific Asia. 3)Fishing category generating the highest profit. 4)Standard shipping mode proving most efficient. 5)Debit card being the preferred payment method. Impact: •This project enabled the prediction of customer-level sales with high accuracy, leading to improved sales forecasting and resource allocation. •Optimized inventory management resulted in reduced stockouts and improved operational efficiency. •Data-driven insights provided valuable guidance for strategic decision-making. Future Work: •Integrate additional data sources (e.g., customer demographics, marketing campaigns) for further model improvement. •Explore advanced machine learning algorithms for potentially higher prediction accuracy. •Implement real-time sales prediction and inventory management system.