This is the winning model from my pharmacy sales forecasting project — a linear regression trained on log-scaled, engineered features. Pick any date from its held-out test period below to see what it predicted, and how confident the model actually was in that specific prediction.

R03 demand predictor

R03 — medication for obstructive airway disease. Pick any date from the held-out test period (2018–2019, never seen during training). The model gives both a point prediction and a per-day confidence interval, computed from how unusual that day's features are relative to what it trained on.

Loading calibration stats…
Loading available range…

Choose a date and hit predict to see the model's forecast and its confidence interval.

Loading model info…

Full write-up: how the model was built · Code: Google Drive