Maps 10–13
Map 10 — Model generalization map
Recall: HISTORY → TRAIN (fit) → VALIDATE (select/tune) → TEST (independent generalization) → DEPLOY; K-FOLD rotates held-out subsets.
Rebuild: add warning arrow: repeated TEST reuse → loss of independence.
Answer check: great train + weak test = over-fitting.
Source: pp. 488–490, 494–495.
Map 11 — Deployment and iteration loop
Recall: USEFUL + FEASIBLE → PRODUCTION → MONITOR VALUE/PERFORMANCE → REFINE → NEW QUESTION/SOURCE → RETRAIN/REEVALUATE.
Answer check: the model is not “done” at deployment.
Source: pp. 490–491.
Map 12 — Governance ring
Recall: SOURCING | SHARING | METADATA | ENRICHMENT | ACCESS | SECURITY/PRIVACY | DQ | VISUALIZATION STANDARDS surround the lifecycle.
Answer check: velocity is not an exception to governance; recombination belongs in security/privacy.
Source: pp. 496–500.
Map 13 — Architecture/tool selection
Recall: PARALLEL ANALYTICS → MPP | FLEXIBLE VARIED-DATA LANDING → DISTRIBUTED FILE | COMPUTE NEAR DATA → IN-DATABASE | ELASTIC/LEASED EXPLORATION → CLOUD.
Meaning: requirement controls tool choice.
Answer check: no platform wins every scenario.
Source: pp. 491–494, 496–497.
Final blank-page test
Rebuild Maps 1, 4, 5, 7, 9, 10, 11, 12 without looking. For each, state one changed fact that moves the decision to another branch.