Iris

Creator and reuse

R. A. Fisher

Review the provider license or terms before reuse.

Before training

Inspect missing values, verify target availability, remove leakage and identifiers, split before fitting transformations, and choose metrics that fit the intended decision.

Data quality checks

Confirm row meaning, duplicates, impossible values, category spelling, target balance, and whether the sample represents the population where the model will be used.

Validation planning

Use random holdout only when rows are independent. Prefer grouped, temporal, or geographic splits when related observations could cross the boundary.