Skip to contents

Project New Rows into a Response-Aligned MFA Space

Usage

# S3 method for class 'response_aligned_mfa'
project(
  x,
  new_data,
  block,
  new_response = NULL,
  response_weight = NULL,
  new_anchor_map = NULL,
  anchor_weight = NULL,
  conditional = FALSE,
  preprocess = TRUE,
  ...
)

Arguments

x

A fitted `response_aligned_mfa` object.

new_data

Numeric matrix/data.frame of new rows for a known predictor block.

block

Character or integer identifying the block used for projection.

new_response

Optional observed multivariate response rows used to refine the latent-score solve when `conditional = TRUE`.

response_weight

Optional scalar or vector of non-negative row weights paired with `new_response`.

new_anchor_map

Optional anchor assignments for the new rows. May be an integer vector of anchor-state ids (with `NA` for unanchored rows) or a non-negative row-stochastic matrix.

anchor_weight

Optional scalar or vector of non-negative row weights paired with `new_anchor_map`.

conditional

Logical; if `TRUE`, allow conditional completion with `new_response`. The default `FALSE` keeps projection pure with respect to the response target.

preprocess

Logical; if `TRUE` (default), applies the fitted predictor and response preprocessing pipelines before projection.

...

Unused.

Value

A numeric matrix of latent scores.