nirs4all.pipeline.storage.chain_replay module

Chain replay – predict from stored chains via WorkspaceStore.

This module provides a standalone replay_chain function that loads a chain from the DuckDB-backed WorkspaceStore and applies it to new feature data.

The function is the primary in-workspace prediction path: it replays preprocessing transformers in order and averages fold-model predictions.

For out-of-workspace prediction, export to .n4a first via WorkspaceStore.export_chain.

Example

>>> from nirs4all.pipeline.storage import WorkspaceStore
>>> from nirs4all.pipeline.storage.chain_replay import replay_chain
>>>
>>> store = WorkspaceStore(workspace_path)
>>> y_pred = replay_chain(store, chain_id="abc123", X=X_new)
nirs4all.pipeline.storage.chain_replay.replay_chain(store: WorkspaceStore, chain_id: str, X: np.ndarray, wavelengths: np.ndarray | None = None) → np.ndarray[source]

Replay a stored chain on new data to produce predictions.

Delegates to WorkspaceStore.replay_chain which loads each step’s artifact, applies the transformation in order, and for the model step loads all fold models and returns the averaged prediction.

Parameters:

  • store – Open WorkspaceStore instance.

  • chain_id – Chain to replay.

  • X – Input feature matrix (n_samples x n_features).

  • wavelengths – Optional wavelength array for wavelength-aware operators.

Returns:

Predicted values as a 1-D numpy.ndarray of shape (n_samples,).

Raises:

  • KeyError – If the chain does not exist.

  • RuntimeError – If the chain has no model step or fold artifacts.