Skip to contents

Flatten Z Array for GLM Input

flatten_Z.Rd

Converts a 4D covariate array into a design matrix compatible with flattened Y. Ensures proper alignment between outcomes and covariates for GLM estimation.

Usage

flatten_Z(Z)

Arguments

Z

Covariate array. Can be:

  • NULL: Returns NULL

  • 3D array (m x m x T): Converted to (m*m*T) x 1 matrix

  • 4D array (m x m x q x T): Converted to (m*m*T) x q matrix

Value

Design matrix with dimensions (m*m*T) x q, or NULL if Z is NULL. Column names are preserved from dimension names or auto-generated.

Details

The flattening process maintains the correspondence between each Y[i,j,t] and its associated covariates Z[i,j,:,t]. The resulting matrix has:

  • Rows: One for each outcome observation (m*m*T total)

  • Columns: One for each covariate (q total)

  • Order: Matches the flattening of Y (column-major)

Special handling for 3D input (single covariate):

  • Automatically detected and converted to column matrix

  • Preserves any dimension names for interpretability