From business problem to ML project (recap of EDA framing)

One-page ML brief per project: decision, target, metric, baseline, constraints, how the prediction is used. Concepts: Problem framing; success criteria; baselines Project A: Metric: business cost 5·FN + FP and Bad recall; baseline: always-Good (cost = 5 × number of Bad); constraint: explainable decisions. Project B: Metric: MAE and peak-hour MAE; baselines: same hour last week and the TSO forecast; constraint: forecast ready by 10:00 the day before.