Prior.coefficient_variance

abstractmethod Prior.coefficient_variance(context)

Prior variance of each coefficient, shaped like the design.

Diagonal by construction. A prior that wants correlated coefficients expresses that through dummy_observations(), where a row can restrict a combination.

Parameters:

context (_PriorContext) – What the prior needs to know about the sample.

Return type:

NDArray[double]

Returns:

A (width, k) array of variances. Entries may be infinite, which is how a prior says it has no opinion about a coefficient.