typing.ForecastFn

A closure that fits a model on a training window and forecasts its test horizon.

Usage

Source

typing.ForecastFn()

Called by backtest() positionally, with full_covariates spanning the full window (train followed by test, i.e. covariates[..., t0:t2, :]) and batch_size forwarded unchanged from backtest so a chunked closure can bound its own device memory. Returns forecast samples with the sample axis first, shape (num_samples, *batch, t2 - t1, obs). The draws may stay in host memory (e.g. via device="host"): a jax Array committed to the CPU backend device or, without a CPU backend, a NumPy array. Every metric in DEFAULT_METRICS accepts such a pred or truth (or both), in any mix and regardless of batch_size, moving a host-resident operand to device memory first where needed; draws already on-device avoid that hop for the metrics scored every window. The parameters are positional-only so a closure keeps its own parameter names. At runtime the beartype hook only checks that the value is callable (Python protocols never inspect signatures); ty checks the signature structurally at the backtest call site.

Methods

Name	Description
__call__()	Fit on the training window and return forecast draws for the test window.

---

__call__()

Fit on the training window and return forecast draws for the test window.

Usage

Source

__call__(
    rng_key,
    model,
    train_data,
    train_covariates,
    full_covariates,
    num_samples,
    /,
    *,
    batch_size=None
)