Member Junction
    Preparing search index...

    Interface SimpleMLModelInfo

    A single trained predictive model from the catalog, surfaced to Interactive Components so they can let users pick a model and weave its predictions into charts/tables. Sourced from the MJ: ML Models entity (one row per immutable, versioned model).

    interface SimpleMLModelInfo {
        holdoutMetrics?: Record<string, unknown>;
        id: string;
        metrics?: Record<string, unknown>;
        pipeline: string;
        problemType: string;
        status: string;
        targetVariable: string;
        version: number;
    }
    Index

    Properties

    holdoutMetrics? id metrics? pipeline problemType status targetVariable version

    Properties

    holdoutMetrics?: Record<string, unknown>

    Parsed holdout metrics scored once on the locked holdout — the honest performance number.

    id: string

    The MJ: ML Models row id — pass this to SimpleMLTools.score.

    metrics?: Record<string, unknown>

    Parsed training metrics, when the model carries them. Shape is algorithm-dependent.

    pipeline: string

    Denormalized name of the training pipeline that produced this model.

    problemType: string

    The kind of prediction the model makes: 'classification' or 'regression'.

    status: string

    Lifecycle status of the model (e.g. 'Published', 'Validated', 'Draft', 'Archived').

    targetVariable: string

    The label/column the model predicts (e.g. "Renewed").

    version: number

    Monotonic model version under the pipeline (higher = newer).

    Settings

    Member Visibility

    On This Page

    Properties