inbound.rotation_warper: Methods for forward (outbound) and backward (inbound) feature...

In alexanderbrenning/wiml: Interpreting machine-learning models in transformed space

Methods for forward (outbound) and backward (inbound) feature space transformation using warper objects

Description

Outbound or forward transformation refers to the mapping T from feature space into a transformed space (e.g., principal component space), and inbound or backward transformation is its inverse transformation T^{1} back into feature space. This is an abstract parent class for pca_warper and other related objects that can be expressed using rotation matrices.

Usage

## S3 method for class 'rotation_warper'
inbound(object, wdata = object$wdata, ...)

## S3 method for class 'rotation_warper'
outbound(object, xdata = NULL, ...)

Arguments

object	A warper object.
wdata	A data frame containing data in transformed space.
...	Currently not used.
xdata	A data frame with features in the original, untransformed feature space.

Details

These functions are mainly intended for internal use. The warp() and unwarp() methods are more user-oriented interfaces.

Value

A transformed data frame of class warped_df (for outbound) or data.frame (for inbound transformations).

Functions

• outbound.rotation_warper: Forward (outbound) transformation using a rotation warper

alexanderbrenning/wiml documentation built on Sept. 29, 2023, 4:45 a.m.