Predictive Stock Level Optimization

I have a full extract of our retail inventory history and I want to turn it into a living model that tells me exactly when, what, and how much to reorder so we stop tying up cash in slow-moving items while never running out of the fast movers. Your task is to dive into the inventory data, uncover the patterns that drive demand, and deliver a predictive engine focused on stock level optimization. Here’s how I picture the engagement: • Data assessment & preparation: explore the raw tables, flag gaps or anomalies, and structure the dataset so the model can consume it without manual fixes each cycle. • Model development: build and tune a demand-driven algorithm (time-series forecasting, probabilistic safety-stock calculations, or a hybrid you prefer) that outputs optimal reorder points and quantities per SKU, factoring seasonality, promotions, and supplier lead times. • Validation & iteration: stress-test accuracy with back-testing, explain any trade-offs between service level and inventory cost, and refine until the metrics hold up. • Deployment package: deliver clean, commented code (Python, R, or equivalent), a concise README, and a simple dashboard or set of visual reports that our planners can refresh with new data. Acceptance criteria 1. Forecast error (MAPE or similar) is clearly reported and beats our current rule-of-thumb approach. 2. Recommended stock levels achieve target service levels we will define together. 3. All code runs end-to-end on our environment with one command. If this sounds like your kind of project, tell me briefly how you would approach the data prep and which modeling technique you believe fits retail inventory best.