Covariance

Covariance(self, dim=1)

Updateable covariance matrix.

Uses Welford’s algorithm to stably update the sample mean and (co)variance estimates.

Methods

Name	Description
covariance	Covariance matrix of the observations.
update	Update mean and (co)variance estimates based on new observations.

covariance

Covariance.covariance(ddof=1)

Covariance matrix of the observations.

Parameters

Name	Type	Description	Default
ddof	int	Delta degrees of freedom (1 for sample covariance, 0 for population)	1

Returns

Name	Type	Description
	np.ndarray	Current covariance matrix estimate of shape (dim, dim)

update

Covariance.update(X)

Update mean and (co)variance estimates based on new observations.

Parameters

Name	Type	Description	Default
X	np.ndarray	(batch_size, dim)-array representing new observations	required