Workflow collection
Forecast evaluation workflow collection
Choose the actual decision you face. Each guide includes an illustrative case, evidence-led tasks and a downloadable manual work brief.
19 guides in this collection. Examples are planning resources; account functionality must be verified separately.
Request TeamBoostAI accessRead signed forecast error without hiding misses
Separate early and late prediction errors from their magnitude.
Before / afterCompare absolute and squared forecast errors
Show how one large miss changes two explicitly defined evaluation measures.
Drafting workbenchScore a toy binary task forecast
Work through a stated probability-error rule without treating one score as calibrated confidence.
Task boardAudit which outcomes fall inside forecast ranges
Record observed interval coverage and width without asserting future coverage from a tiny sample.
Readiness checklistKeep unresolved outcomes outside scored verdicts
Preserve missing evaluation labels rather than counting unknown results as failed forecasts.
Decision deskMatch a forecast to its decision horizon
Distinguish next-item timing from a whole batch or milestone forecast.
Handoff sequencePreserve the forecast made before the outcome
Prevent a revised post-outcome number from replacing the prediction being evaluated.
Before / afterCombine estimates with shared scenario assumptions
Use coherent joint scenarios rather than mixing incompatible optimistic and pessimistic cases.
Manager memoSeparate a planning range from a confidence claim
Keep a scenario envelope visibly different from a calibrated probability interval.
Readiness checklistCheck a forecast's work-class eligibility
Restrict evaluation to the task class and decision context that the model actually describes.
Troubleshooting pathKeep a zero baseline out of a scaled-error claim
Detect an undefined relative forecast score when the training reference has no variation.
Decision deskInspect a forecast decision with asymmetric mistake costs
Compare the supplied losses of overcommitting and waiting before using a probability threshold.
Handoff sequencePlan a rolling-origin task forecast evaluation
Keep each held-out outcome beyond its own fitting cutoff while matching a fixed forecast horizon.
Troubleshooting pathKeep zero outcomes out of undefined percentage-error claims
Inspect a forecast error when its observed-value denominator is zero.
Readiness checklistFreeze a forecast evaluation cutoff before fitting
Separate held-out task outcomes from evidence used to choose the forecast rule.
Before / afterCompare forecast methods on matched cases
Compare forecast methods on the same observed cases while retaining missing predictions.
Drafting workbenchCompare a task classifier with a simple base-rate rule
Check whether high accuracy hides failure to recognize the less frequent required outcome.
Drafting workbenchInspect forecast probabilities by relevant work group
Retain conditional group results when an aggregate probability comparison hides different behavior.
Acceptance rubricInspect false positives and false negatives at a decision threshold
Keep the consequences of a binary work-readiness classification visible beside accuracy.