SVDTransformer

class lightautoml.transformers.decomposition.SVDTransformer(subs=None, random_state=42, n_components=100)[source]

Bases: LAMLTransformer

TruncatedSVD.

Parameters:

  • subs (Optional[int]) – Subsample to fit algorithm. If None - full data.

  • random_state (int) – Random state to take subsample.

  • n_components (int) – Number of SVD components.

property features

Features list.

fit(dataset)[source]

Fit algorithm on dataset.

Parameters:

dataset (Union[NumpyDataset, CSRSparseDataset]) – Sparse or Numpy dataset of text features.

Returns:

self.

transform(dataset)[source]

Transform input dataset to SVD representation.

Parameters:

dataset (Union[NumpyDataset, CSRSparseDataset]) – Sparse or Numpy dataset of text features.

Return type:

NumpyDataset

Returns:

Numpy dataset with text embeddings.