[D] Building a demand forecasting system for multi-location retail with no POS integration, architecture feedback wanted

About this article

We’re building a lightweight demand forecasting engine on top of manually entered operational data. No POS integration, no external feeds. Deliberately constrained by design. The setup: operators log 4 to 5 signals daily (revenue, covers, waste, category mix, contextual flags like weather or local events). The engine outputs a weekly forward-looking directive. What to expect, what to prep, what to order. With a stated confidence level. Current architecture thinking: Days 1 to 30: statistical ...