Dataset Shift

Inference, estimation, and decision-making from data

Most evaluation methods quietly assume that your training examples, test examples, and future examples all come from the same stable process. That assumption has a name: independent and identically distributed, or i.i.d. It says every example is drawn independently from one fixed distribution, and that distribution never changes.

🔒 This is a Pro lesson — the interactive figure, worked examples, quiz and practice open with Pro access.

▶ Dataset Shift
← Cross-ValidationEvaluation Metrics →