AI Trend & Stock Prediction Module

I need an AI-powered module added to my existing e-commerce site to forecast customer demand and recommend optimal stock levels for the household goods we sell. All historical sales figures are already captured daily and exported as Excel/CSV files, so the model can start from a clean, structured dataset right away.

Here’s the flow I imagine: the script ingests the daily sales files, trains or updates the model, and then surfaces two clear outputs on the admin dashboard—a rolling demand forecast (7, 14 and 30-day windows) and automatic restock suggestions per SKU. I want these insights to refresh automatically each night after the day’s file is saved, with a manual “re-run” button for ad-hoc checks.

The site is built in PHP with a MySQL backend, but I’m fine if the predictive engine itself runs in Python using pandas, scikit-learn or TensorFlow and simply exposes results back to PHP through an API endpoint or scheduled cron job.

Deliverables will be:
• A working prediction script or micro-service linked to the current site
• One-click or scheduled retraining tied to the daily CSV upload
• A lightweight admin panel widget that displays demand curves and restock quantities in a readable format
• Brief documentation outlining model assumptions, how to tune parameters, and how to extend the time horizon

When I can log in, trigger a forecast, and see accurate suggestions that match recent sales patterns, I’ll consider the job complete.