dlm-js
    Preparing search index...

    Interface DlmForecastOptions

    Options for dlmForecast.

    interface DlmForecastOptions {
        dtype?: DlmDtype;
        obsStd?: number;
        timestamps?: number[];
        X?: ArrayLike<number>[];
    }

    Properties

    dtype?: DlmDtype

    Computation precision (should match dtype used in dlmFit). Default: 'f64'.

    obsStd?: number

    Observation noise std dev (scalar). Overrides the value from the fit result. If omitted, defaults to fit.obsNoise[0] (the first observation's noise std dev). Useful for scenario analysis ("what if observation noise were different?").

    timestamps?: number[]

    Timestamps for forecast steps (length h). When provided, forecast-step intervals Δt_k = timestamps[k] - timestamps[k-1] (or timestamps[0] - last fit timestamp) are used to compute time-varying G(Δt_k) and W(Δt_k). When omitted, all forecast steps use Δt = 1 (uniform spacing).

    X?: ArrayLike<number>[]

    Covariate rows for forecast steps (h rows × q cols).

    Properties