Tune Bilinear Mixed Hyperparameters via Subject-Block CV

Tunes a compact set of key parameters (`r_seed`, `r_roi`, `k_subject`, optional `lambda_y`) using subject-blocked cross-validation.

Usage

bilinear_mixed_tune(
  data,
  subject,
  z = NULL,
  y = NULL,
  row_design = NULL,
  mode = c("auto", "seed_axis", "seed_repeat", "both"),
  connectivity_type = c("auto", "cross", "symmetric"),
  profile = c("fast", "balanced", "adaptive"),
  grid = NULL,
  n_folds = 3,
  metric = c("auto", "reconstruction", "trait_r2"),
  seed = 1,
  verbose = FALSE,
  ...
)

Arguments

data

A list of numeric connectivity matrices.

subject

Subject identifier.

z

Optional repeat-level design.

y

Optional subject-level traits.

row_design

Optional row-level covariates.

mode

One of "auto", "seed_axis", "seed_repeat", or "both".

connectivity_type

One of "auto", "cross", or "symmetric".

profile

One of "fast", "balanced", or "adaptive".

grid

Optional candidate grid. Either:

• a data.frame with columns among r_seed, r_roi, k_subject, lambda_y

• or a named list of vectors for those fields.

n_folds

Number of subject-block CV folds.

metric

One of "auto", "reconstruction", or "trait_r2".

seed

Random seed for fold assignment.

verbose

Logical verbosity.

...

Additional fixed arguments forwarded to [bilinear_mixed()].

Value

A list of class "bilinear_mixed_tuning" with best_params, results, and refit fit.