Skip to contents

Cluster-to-latent loadings for DKGE subjects

Source: R/dkge-latent-utils.R
dkge_cluster_loadings.Rd

Computes \(A_s = B_s^\top K U\), the linear map that pulls latent-space vectors back to subject cluster space. Multiplying \(A_s\) by a latent coefficient vector produces a Haufe-style decoder at the subject level.

Usage

dkge_cluster_loadings(fit)

Arguments

fit

Fitted object of class `dkge` containing `Btil`, `K`, and `U`.

Value

A list of `P_s x r` matrices of cluster loadings.