Sales Forecasting

Modeled daily and weekly unit demand from order history, calendar effects, and known commercial events.

Gave planning a documented forecast instead of a trailing average, and fed the executive review described in the scorecard case study.

Role: Designed the forecasting workflow and compared models against held-out history.
Scale: Daily and weekly unit sales for e-commerce planning.

Approach

I aggregated order history to daily and weekly units and added calendar features, holiday flags, and known pricing events. Models were compared on held-out periods. The useful model was the one a reviewer could question, not the one with the most machinery.

This work is part of the Executive Scorecard & Forecasting story. It is listed here so the modeling workstream stays findable.