metrics

---

iso_doc_type: "Specification" iso_viewpoint: "ComponentView" type: "module" title: "regression_model_template::core::metrics — Metric Evaluation Abstractions" source_path: "src/regression_model_template/core/metrics.py" description: "Abstract Metric base class, Scikit-Learn metric wrapper, and MLflow threshold evaluators." tags: ["iso42010", "okf", "component_view", "metrics", "sklearn"] timestamp: "2026-10-01T17:50:00Z" generated: "agent:wiki-iso-documentation-standard" verified: "true" last_verified_commit: "073b5fd"

regression_model_template::core::metrics — Metric Evaluation Abstractions

Source: src/regression_model_template/core/metrics.py (Lines: L1-L148)
Layer: Domain
Role: Encapsulates regression evaluation scoring functions, threshold criteria, and MLflow serialization.

---

1. Class Diagram

classDiagram
    class Metric {
        <<abstract>>
        +str KIND
        +str name
        +bool greater_is_better
        +score(targets: Targets, outputs: Outputs) float*
        +scorer(model: Model, inputs: Inputs, targets: Targets) float
        +to_mlflow() MlflowMetric
    }

    class SklearnMetric {
        +str KIND = "SklearnMetric"
        +str name
        +bool greater_is_better
        +score(targets: Targets, outputs: Outputs) float
    }
    SklearnMetric --|> Metric : implements

    class Threshold {
        +float threshold
        +bool greater_is_better
        +to_mlflow() MlflowThreshold
    }

Loading

---

2. Method Contracts

Metric.score(self, targets: schemas.Targets, outputs: schemas.Outputs) -> float

• Source Citation: src/regression_model_template/core/metrics.py:L44-L53
• Visibility: Public (+)
• Behavior: Evaluates predictive quality between true targets and model predictions.

---

Related: Models · Evaluations Job