Create an MVPA Dataset for Clustered/Parcellated Data

Creates a dataset object for MVPA analysis on parcellated brain data stored as ClusteredNeuroVec objects from neuroim2. A "clustered searchlight" iterates over all K parcels, pooling nearby parcels as features for each analysis.

Usage

mvpa_clustered_dataset(train_data, test_data = NULL, mask = NULL)

Arguments

train_data

A ClusteredNeuroVec instance (training data).

test_data

An optional ClusteredNeuroVec instance (test data). Default: NULL.

mask

An optional mask. If NULL, derived from train_data's cluster volume.

Value

An mvpa_clustered_dataset object (S3 class) containing:

train_data

The training data as a ClusteredNeuroVec.

test_data

The test data (if provided).

mask

The mask.

cvol

The ClusteredNeuroVol defining cluster assignments.

region_mask

A NeuroVol where each voxel's value is its cluster ID (for broadcasting results via map_values).

has_test_set

Logical flag.

Examples

ds <- gen_clustered_sample_dataset(D = c(5, 5, 5), nobs = 20, K = 5)
print(ds$dataset)
#> 
#>   Clustered MVPA Dataset
#> 
#>   - Clusters:      5 
#>   - Observations:  20 
#>   - Test set:      no 
#>