beginnerBaselines

Beat "Same as Last Week"

Decide what you would do from the brief alone, including whether you would change anything at all. Everything below it is available, but the exercise stops working if you open it first.

The brief you were given

The sales-ops team asks for a weekly demand forecast per SKU "using machine learning, because the spreadsheet is just last week's number". They want it in a month. A first model exists: a gradient-boosted regressor on calendar and price features.

The trap — the fix that moves the metric and is not the fix

Tuning the boosted model until it beats the spreadsheet's error on the random split, then shipping it everywhere. The comparison chart is convincing, the sales-ops team gets what they asked for, and the random split has let the model see the week after each week it forecasts. In production the stable SKUs get a noisier forecast than they had, the ordering team notices within a month, and the conclusion drawn is that machine learning does not work for demand — when the honest result was that it worked for the 15% of weeks that mattered.

Read this even if you are confident. It is here rather than behind a button because it is the answer most teams actually ship, it passes review, and its cost arrives weeks later when the labels do.