Kelvin-N-Analytics/beer-demand-forecasting-east-africa

End-to-end supply chain analytics project: demand forecasting, inventory optimization and supplier analysis for an East African FMCG beverage company | Python · Pandas · Matplotlib # 🍺 Beer Demand Forecasting & Inventory Optimization East Africa FMCG > **A end-to-end supply chain analytics project simulating real-world demand planning and inventory optimization for a beverage company operating across Kenya.** --- ## 📌 Project Overview FMCG companies in East Africa face a common challenge: **too much stock of the wrong product, in the wrong place, at the wrong time.** This results in high holding costs, stockouts during peak seasons, and reactive supply chain decisions. This project simulates 3 years of beer and spirits sales data across 4 Kenyan regions and 6 SKUs, then applies demand forecasting and inventory optimization techniques to quantify how much better analytical decisions can save a business. **Target companies this analysis mirrors:** East African Breweries Limited (EABL) and Kenya Wine Agencies Limited (KWAL). --- ## 🎯 Business Questions Answered 1. What does demand look like across regions and products and when does it spike? 2. How much does running a promotion actually lift sales? 3. Can we forecast demand more accurately than a simple moving average? 4. What is the optimal safety stock and reorder point for each SKU? 5. How much money does better forecasting save in holding costs? 6. Which suppliers and regions have the most lead time risk? --- ## 📊 Dataset | Parameter | Detail | |---|---| | Records | 864 rows | | Products | 6 SKUs (Tusker Lager, Senator Keg, Tusker Malt, Johnnie Walker Black, Smirnoff Vodka, Baileys Original) | | Regions | Nairobi, Mombasa, Kisumu, Nakuru | | Time Period | January 2022 – December 2024 (36 months) | | Features | Demand, Opening/Closing Stock, Lead Time, Promotions, Unit Price, Revenue | > **Note:** Data is simulated using realistic FMCG assumptions including seasonality, regional demand multipliers, promotional uplifts, and random variation. It is designed to mirror the data environment at EABL/KWAL. --- ## 🔍 Key Findings ### 1. Seasonality & Demand Patterns - **December** is …