mojolearn.QuadraticDiscriminantAnalysis

class mojolearn.QuadraticDiscriminantAnalysis(*, solver='svd', shrinkage=None, priors=None, reg_param=0.0, store_covariance=False, tol=0.0001, covariance_estimator=None)

Bases: _Classifier

Quadratic discriminant analysis classifier.

Follows scikit-learn’s QuadraticDiscriminantAnalysis. It fits one Gaussian per class, each with its own covariance, and classifies by the resulting quadratic decision function. Computation is float32.

Parameters:

• solver ({‘svd’, ‘eigen’}, default ‘svd’) – 'svd' regularizes with reg_param. 'eigen' uses the empirical (optionally shrunk) class covariances and ignores reg_param, as in scikit-learn.

• shrinkage ({None, ‘auto’} or float, default None) – Covariance shrinkage for 'eigen': 'auto' uses the Ledoit-Wolf estimate, a float in [0, 1] sets it directly. With solver='svd' it raises NotImplementedError.

• priors (array-like of shape (n_classes,), default None) – Class priors (nonnegative). None uses the class frequencies.

• reg_param (float, default 0.0) – Regularization of the class covariances under 'svd'.

• store_covariance (bool, default False) – Store each class covariance in covariance_.

• tol (float, default 1e-4) – A class whose scalings are not all above tol is rejected with ValueError (its covariance is not full rank).

• covariance_estimator (object, default None) – An estimator with fit and a covariance_ attribute, used instead of the empirical covariance with solver='eigen' (ValueError with 'svd' or with shrinkage set). It is fitted on the host on each class’s float32 rows.

Variables:

• classes_ (array of shape (n_classes,)) – Class labels.

• means_ (array of shape (n_classes, n_features)) – Class means.

• priors_ (array of shape (n_classes,)) – Class priors used.

• rotations_ (list of arrays) – Per class, the (n_features, n_features) eigenvectors of its covariance.

• scalings_ (list of arrays) – Per class, the (n_features,) variances along those eigenvectors.

• covariance_ (list of arrays) – Per class covariance, when store_covariance=True.

• n_features_in_ (int) – Number of columns seen in fit.

• numeric_mode_ (str) – Numeric mode the estimator was fitted in; prediction runs in the same mode.

Notes

Fit raises ValueError with fewer than two classes or when a class has a single row.

Examples

>>> import numpy as np, mojolearn
>>> rng = np.random.default_rng(0)
>>> X = rng.random((1000, 4), dtype=np.float32)
>>> y = rng.integers(0, 2, 1000)
>>> model = mojolearn.QuadraticDiscriminantAnalysis().fit(X, y)

__init__(*, solver='svd', shrinkage=None, priors=None, reg_param=0.0, store_covariance=False, tol=0.0001, covariance_estimator=None)

Methods

decision_function(X)
fit(X, y)
fit_transform(X[, y])	Fit to X, then transform X.
get_params([deep])	Return the constructor parameters as a dict.
predict(X)	Predict the class of each row.
predict_joint_log_proba(X)	Return the unnormalized log score of each row and class.
predict_log_proba(X)	Return the log of the class probabilities for each row.
predict_proba(X)	Return class probabilities for each row.
score(X, y)	Return the mean accuracy on the given rows and labels.
set_params(**params)	Set constructor parameters.