Partially project data for a cross_projector — partial_project.cross

Projects new data from either the X or Y domain onto the latent subspace, considering only a specified subset of original features (colind).

Usage

# S3 method for class 'cross_projector'
partial_project(
  x,
  new_data,
  colind,
  least_squares = TRUE,
  lambda = 1e-06,
  source = c("X", "Y"),
  ...
)

Arguments

x: A cross_projector object.
new_data: A numeric matrix (n x length(colind)) or vector, representing the observations corresponding to the columns specified by colind.
colind: A numeric vector of column indices in the original data space (either X or Y domain, specified by source) that correspond to new_data's columns.
least_squares: Logical; if TRUE (default), use ridge-regularized least squares for projection.
lambda: Numeric; ridge penalty (default 1e-6). Ignored if least_squares=FALSE.
source: Character, either "X" or "Y", indicating which domain new_data and colind belong to.
...: Additional arguments (currently ignored).

Value

A numeric matrix (n x d) of factor scores in the latent subspace.