Sample stats
Dataset composition and search-key quality used for FIT evaluation.
Monthly count and mean target
Samples may overlap in the same calendar month · statistics use each sample as provided
Performance
Baseline and enriched quality dynamics for the selected sample. 5-fold validation · comparable labeled rows · 95% CI.
BaselineEnrichedUplift
Score analysis
Mean score and mean target on the 0–1 scale; labeled and unlabeled counts on the secondary axis.
Score distribution by target
Normalized probability of scores by common bins. Unlabeled rows are excluded.
target = 0target = 1
Score stability (PSI)
Monthly PSI of scores · reference: first period (baseline month) · fixed reference bins.
Features SHAP
Global mean absolute SHAP for features used by the evaluated model.
Search results
Relevant features, data sources, and generated AutoFE descriptions.
| Feature | Provider | Source | SHAP | PSI | Drift | Coverage |
|---|
No features match the selected filters.
| Provider | Source | Aggregate SHAP | Relevant features | Action |
|---|
| Sources | Generated feature | Source features | Functions |
|---|