πΏπ² Streamlit web application for Zambia national retail analytics β live BI dashboard, demand forecasting (XGBoost & ARIMA), customer segmentation (RFM + K-Means) & stockout risk detection. Deployed on Streamlit Community Cloud.
zambia-retail-dashboard-appβ¦
# πΏπ² Zambia National Retail Intelligence Platform
**A full-stack retail analytics dashboard for the Zambian market β featuring demand forecasting, customer segmentation, risk detection, and executive reporting, all powered by a synthetic 80,000-record dataset.**
Features Β· Screenshots Β· Installation Β· Usage Β· Architecture Β· Data Model Β· ML Models Β· Roadmap
---
## π Overview
The **Zambia National Retail Intelligence Platform** is an interactive business intelligence web application built with Streamlit. It simulates and analyses retail transaction data across Zambia's major cities and store chains from **2022 to 2024**, delivering actionable insights across six analytical modules:
zambia-retail-dashboard-appβ¦
- Executive KPI dashboards with real-time filtering
- Exploratory data analysis with seasonality and promotion impact
- ML-powered demand forecasting (XGBoost & ARIMA)
- RFM-based customer segmentation using K-Means clustering
- Stockout and anomaly risk detection
- Interactive raw data explorer
> **Note:** All data in this application is **synthetically generated** for demonstration and analytical purposes. It does not represent real commercial data from any Zambian retailer.
---
## β¨ Features
### π Executive Dashboard
- Seven headline KPI cards: Total Revenue, Gross Profit, Average Margin, Transactions, Customers, Stockout Rate, and Promotion Rate
- Revenue breakdown by city, month, category, product, and store type
- Year-on-year revenue comparison
- Gross margin analysis by product category
### π EDA & Trends
- Seasonality analysis β average revenue by month and day of week
- Promotion impact analysis: revenue lift %, margin trade-off comparison
- City and regional revenue breakdown with stacked bar charts
- Revenue treemap drill-down: City β Store
### π Demand Forecasting
Two selectable forecasting models:
| Model | D β¦