mojolearn.TruncatedSVD#

class mojolearn.TruncatedSVD(n_components=2, *, algorithm='covariance_eigh', n_iter=5, n_oversamples=10, power_iteration_normalizer='auto', random_state=None, tol=0.0)[source]#

Bases: NumericModeMixin

Dimensionality reduction by truncated singular value decomposition.

Unlike PCA, the data is not centered, so this works on raw term counts or TF-IDF matrices (a scipy.sparse input is densified). Input is converted to C-order float32. The default algorithm is the exact "covariance_eigh" solver, not scikit-learn’s "randomized".

Parameters:

n_components (int, default 2) – Number of components, in [1, n_features].
algorithm ({“covariance_eigh”, “jacobi”, “randomized”}, default “covariance_eigh”) – "covariance_eigh" and "jacobi" run the exact solver (an eigendecomposition of the Gram matrix). "randomized" runs randomized_svd(). "arpack" and any other value raise NotImplementedError.
n_iter (int, default 5) – Power iterations for "randomized".
n_oversamples (int, default 10) – Extra random vectors for "randomized".
power_iteration_normalizer (str, default “auto”) – Normalizer between power iterations for "randomized".
random_state (int or None, default None) – Seed for "randomized".
tol (float, default 0.0) – Only 0.0 is accepted (it is ARPACK’s tolerance); any other value raises NotImplementedError.

Variables:

components_ (Array of shape (n_components, n_features), float32) – Right singular vectors.
singular_values_ (Array of shape (n_components,), float32) – Singular values.
explained_variance_ (Array of shape (n_components,), float32) – Variance of each transformed column (scikit-learn’s definition).
explained_variance_ratio_ (Array of shape (n_components,), float32) – explained_variance_ divided by the total variance of X.
n_components_ (int) – Number of components.
n_features_in_ (int) – Number of features seen during fit.
input_copied_ (bool) – Whether X had to be copied to C-order float32.

Examples

>>> import numpy as np, mojolearn
>>> X = np.random.default_rng(0).random((1000, 8), dtype=np.float32)
>>> svd = mojolearn.TruncatedSVD(n_components=3).fit(X)
>>> Z = svd.transform(X)

__init__(n_components=2, *, algorithm='covariance_eigh', n_iter=5, n_oversamples=10, power_iteration_normalizer='auto', random_state=None, tol=0.0)[source]#

Methods

`fit`(X[, y])	Fit the model to `X`.
`fit_transform`(X[, y])	Fit to `X` and return its projection.
`get_params`([deep])
`inverse_transform`(X)	Map reduced data back to the original feature space.
`load`(path)	Load a model written by `save`.
`numeric_mode_used`()	The numeric mode this estimator runs in.
`save`(path)	Save the fitted model to `path` as an `.npz` file.
`set_params`(**params)	Set constructor parameters.
`transform`(X)	Project `X` onto the components.
`vendor_used`()	The GPU backend this estimator runs on.

fit(X, y=None)[source]#

Fit the model to X.

Parameters:

X (array-like of shape (n_samples, n_features)) – Training data; at least 2 rows and 2 features.
y (ignored)

Returns:

self (TruncatedSVD) – The fitted estimator.

Raises:

ValueError – If X is too small or n_components is out of range.
NotImplementedError – For an unsupported algorithm or a nonzero tol.

transform(X)[source]#

Project X onto the components.

Parameters:: X (array-like of shape (n_samples, n_features)) – Data with the training feature count.
Returns:: Z (Array of shape (n_samples, n_components), float32) – The reduced data.
Raises:: ValueError – If the model is not fitted or the feature count differs.

fit_transform(X, y=None)[source]#

Fit to X and return its projection.

Parameters:

X (array-like of shape (n_samples, n_features)) – Training data.
y (ignored)

Returns:

Z (Array of shape (n_samples, n_components), float32) – The reduced data.

inverse_transform(X)[source]#

Map reduced data back to the original feature space.

Parameters:: X (array-like of shape (n_samples, n_components)) – Reduced data, as returned by transform.
Returns:: X_original (Array of shape (n_samples, n_features), float32) – The reconstruction X @ components_.
Raises:: ValueError – If the component count differs from the fit.

save(path)[source]#

Save the fitted model to path as an .npz file.

A model restored with TruncatedSVD.load or mojolearn.host_model(path) transforms, including on a CPU-only install.

Parameters:: path (str or path-like) – Destination file.
Raises:: RuntimeError – If the model is not fitted.

classmethod load(path)[source]#

Load a model written by save.

The loaded model transforms; it does not refit.

Parameters:: path (str or path-like) – File written by TruncatedSVD.save.
Returns:: model (TruncatedSVD) – The restored model.
Raises:: ValueError – If the file is not a TruncatedSVD model or its arrays are inconsistent.