This project provides a complete, end-to-end analytics platform for retail decision-making. It integrates product-level forecasting, customer segmentation, and actionable insights into a unified Python-based workflow.
The system helps retailers make informed decisions related to demand forecasting, inventory planning, customer retention, and product strategy. It combines machine learning, time-series modeling, and RFM-based CRM analytics to deliver reliable forecasts and meaningful insights.
- Removes negative or invalid values
- Standardizes column names
- Fixes missing data
- Computes TotalPrice automatically
- Applies weekly aggregation for forecasting stability
The system supports two forecasting models:
- SARIMA for long-term, trend-aware forecasting
- XGBoost for short-term, high-accuracy predictions
Both forecasts are displayed on the same graph when a product is entered, supporting manufacturing and inventory planning.
Customers are categorized using Recency, Frequency, and Monetary scoring:
- Top Spenders
- New Customers
- At-Risk / Dormant
Each group automatically receives tailored discount recommendations:
- 15% VIP Discount (Top Spenders)
- 5% Welcome Offer (New Customers)
- 10% Winback Coupon (At-Risk)
Retailers can search any Customer ID to view:
- Complete RFM profile
- Purchase history
- Total spend and quantity
- Most purchased items
The system identifies:
- Top 20 best-selling products
- Bottom 20 worst-performing products
Metrics include total sales, quantity sold, and unique customers.
- Python (pandas, numpy, matplotlib)
- Statsmodels (SARIMA)
- XGBoost
- Scikit-learn
- OpenPyXL
- IPython Display
Enter a product name or stock code to generate forecasts.
Enter a Customer ID to view their RFM profile and purchase history.
This system enables retail businesses to understand customer behavior, anticipate product demand, and optimize stocking and marketing decisions through a unified predictive analytics platform.