ddl.linear.LinearProjector

class ddl.linear.LinearProjector(linear_estimator=None, orthogonal=False, fit_bias=True)

Bases: sklearn.base.BaseEstimator, ddl.base.ScoreMixin, sklearn.base.TransformerMixin

A linear projector based on an underlying linear estimator.

(Somewhat like a relative destructor with an implicit but unknown underlying density.)

A few notes:

1. By construction, LinearProjector is density-agnostic and therefore is not a valid destructor. 2. However, if attached before any valid destructor, the joint transformer is a valid destructor, which implies a valid joint density. 3. Thus, LinearProjector can be seen as a relative destructor that requires a base density to be useful.

Parameters:

linear_estimator : estimator, default=IdentityLinearEstimator

A linear estimator that has either a coef_ attribute or a components_. For example, sklearn.decomposition.PCA.

orthogonal : bool, default=False

Whether to issue a warning if the matrix is not orthogonal.

fit_bias : bool, default=True

Whether to fit the bias term, i.e., the b of the linear transform y = Ax + b.

Attributes:

A_ : object

Representation of linear transformation, i.e. a matrix. The linear object that implements dot(X), logabsdet, and inv. Note that this can be a compact representation of a matrix such as merely storing the diagonal elements or a Householder transformation matrix.

A_inv_ :

Representation of the inverse linear transformation, i.e. a matrix. The linear object that implements dot(X), logabsdet, and inv. Note that this can be a compact representation of a matrix such as merely storing the diagonal elements or a Householder transformation matrix.

Methods

create_fitted([A, b, coef, intercept, …])	Create fitted linear projector.
fit(self, X[, y, lin_est_fit_params, …])	[Placeholder].
fit_transform(self, X[, y])	Fit to data, then transform it.
get_domain(self)	Get the domain of this destructor.
get_params(self[, deep])	Get parameters for this estimator.
inverse_transform(self, X[, y])	Apply inverse destructive transformation to X.
score(self, X[, y])	Return the mean log likelihood (or log(det(Jacobian))).
score_samples(self, X[, y])	Compute log(det(Jacobian)) for each sample.
set_params(self, \*\*params)	Set the parameters of this estimator.
transform(self, X[, y])	Apply destructive transformation to X.

__init__(self, linear_estimator=None, orthogonal=False, fit_bias=True)

Initialize self. See help(type(self)) for accurate signature.

classmethod create_fitted(A=None, b=None, coef=None, intercept=None, fitted_linear_estimator=None, **kwargs)

Create fitted linear projector.

Must provide (A and b) OR (coef and intercept) OR (fitted_linear_estimator) but only one combination should be provided.

Parameters:

A : array-like, shape (n_features, n_features), optional

Matrix for projection.

b : array-like, shape (n_features,), optional

Shift vector for linear projection.

coef : array-like, shape (n_features,), optional

Vector of coefficents, will form scaled Householder reflector.

intercept : float, optional

Intecept for 1D linear projection.

fitted_linear_estimator : fitted estimator, optional

Must be a linear estimator that has coef_ or components_ fitted parameters.

**kwargs

Other parameters to pass to constructor.

Returns:

fitted_transformer : Transformer

Fitted transformer.

fit(self, X, y=None, lin_est_fit_params=None, fitted_lin_est=None)

[Placeholder].

Parameters:

X :

y :

lin_est_fit_params :

fitted_lin_est :

Returns:

obj : object

fit_transform(self, X, y=None, **fit_params)

Fit to data, then transform it.

Fits transformer to X and y with optional parameters fit_params and returns a transformed version of X.

Parameters:

X : numpy array of shape [n_samples, n_features]

Training set.

y : numpy array of shape [n_samples]

Target values.

Returns:

X_new : numpy array of shape [n_samples, n_features_new]

Transformed array.

get_domain(self)

Get the domain of this destructor.

Returns:

domain : array-like, shape (2,) or shape (n_features, 2)

If shape is (2, ), then domain[0] is the minimum and domain[1] is the maximum for all features. If shape is (n_features, 2), then each feature’s domain (which could be different for each feature) is given similar to the first case.

get_params(self, deep=True)

Get parameters for this estimator.

Parameters:

deep : boolean, optional

If True, will return the parameters for this estimator and contained subobjects that are estimators.

Returns:

params : mapping of string to any

Parameter names mapped to their values.

inverse_transform(self, X, y=None)

Apply inverse destructive transformation to X.

Parameters:

X : array-like, shape (n_samples, n_features)

New data, where n_samples is the number of samples and n_features is the number of features.

y : None, default=None

Not used in the transformation but kept for compatibility.

Returns:

X_new : array-like, shape (n_samples, n_features)

Transformed data.

score(self, X, y=None)

Return the mean log likelihood (or log(det(Jacobian))).

Parameters:

X : array-like, shape (n_samples, n_features)

New data, where n_samples is the number of samples and n_features is the number of features.

y : None, default=None

Not used but kept for compatibility.

Returns:

log_likelihood : float

Mean log likelihood data points in X.

score_samples(self, X, y=None)

Compute log(det(Jacobian)) for each sample.

Note that this is not the log-likelihood since this is not a destructor. This score is constant for all samples because the Jacobian is constant for linear operations and depends only on the transformation matrix.

Parameters:

X : array-like, shape (n_samples, n_features)

New data, where n_samples is the number of samples and n_features is the number of features.

y : None, default=None

Not used but kept for compatibility.

Returns:

log_det_jacobian : array, shape (n_samples,)

Log determinant of the Jacobian for each data point in X.

set_params(self, **params)

Set the parameters of this estimator.

The method works on simple estimators as well as on nested objects (such as pipelines). The latter have parameters of the form <component>__<parameter> so that it’s possible to update each component of a nested object.

Returns:

self

transform(self, X, y=None)

Apply destructive transformation to X.

Parameters:

X : array-like, shape (n_samples, n_features)

New data, where n_samples is the number of samples and n_features is the number of features.

y : None, default=None

Not used in the transformation but kept for compatibility.

Returns:

X_new : array-like, shape (n_samples, n_features)

Transformed data.