transform
transform
Fitted transformations applied at the model-training boundary.
Transforms must fit only on fit partitions and retain that fitted state for validation, testing, and inference. They are not globally materialized market features.
Classes
| Name | Description |
|---|---|
| Pipeline | Wrap sklearn’s pipeline with QRT split-aware fitting. |
Pipeline
transform.Pipeline(steps)Wrap sklearn’s pipeline with QRT split-aware fitting.
Parameters
| Name | Type | Description | Default |
|---|---|---|---|
| steps | Sequence[tuple[str, Any]] | Sklearn-style (name, transformer) pairs. |
required |
Learned state is fitted only from partitions whose role is "fit". The fitted sklearn pipeline then transforms the complete aligned feature frame, retaining targets, weights, metadata, and split schemes.
Methods
| Name | Description |
|---|---|
| fit | Fit sklearn transformers using only role-fit partitions. |
| fit_transform | Fit on allowed rows, then transform the complete dataset. |
| transform | Transform every row and preserve all non-feature components. |
fit
transform.Pipeline.fit(dataset, *, scheme=None, fold=None)Fit sklearn transformers using only role-fit partitions.
fit_transform
transform.Pipeline.fit_transform(dataset, *, scheme=None, fold=None)Fit on allowed rows, then transform the complete dataset.
transform
transform.Pipeline.transform(dataset)Transform every row and preserve all non-feature components.