Mixed-effect multivariate regression

Fit the row-side design geometry for operator-valued ANOVA. Supports fixed-effect designs (random = NULL) and grouped random-effects designs with a single grouping variable (random intercept and optional random slopes, shared-Omega across response features).

Usage

mixed_regress(
  Y,
  design = NULL,
  fixed,
  random = NULL,
  basis = identity_basis(),
  preproc = center(),
  term_scopes = NULL,
  exchangeability = NULL,
  ...
)

Arguments

Y

Response matrix (n_obs x p) or 3D array (n_subject x n_within x p).

design

Optional design data frame. Required for matrix input.

fixed

Fixed-effect formula.

random

Random-effect specification. Either NULL (fixed-effects only) or a one-sided formula of the form ~ ... | group understood by reformulas::findbars() / lme4::findbars(). All random-effects bars must share a single grouping variable.

basis

Feature basis specification or projector-like object.

preproc

Response preprocessor.

term_scopes

Optional named character vector or named list overriding automatic effect-scope classification for fixed-effect terms. Allowed values are "between", "within", "mixed", and "ungrouped".

exchangeability

Optional named character vector or named list specifying permutation/block-resampling structure for fixed-effect terms. Allowed values are "between_subject", "within_subject", "whole_subject", and "rows". When omitted, exchangeability is inferred from grouped-design scope.

...

Reserved for future extensions.

Value

A mixed_fit object.