tflr: Transformation-free linear regression (TFLR) for...

In Compositional: Compositional Data Analysis

Transformation-free linear regression (TFLR) for compositional responses and predictors

Description

Transformation-free linear regression (TFLR) for compositional responses and predictors.

Usage

tflr(y, x, tol = 1e-6, xnew = NULL)
tflr.irls(y, x, xnew = NULL, tol = 1e-08, maxit = 100)

Arguments

y	A matrix with the compositional response. Zero values are allowed.
x	A matrix with the compositional predictors. Zero values are in general allowed, but there can be cases when these are problematic.
xnew	If you have new data use it, otherwise leave it NULL.
tol	The tolerance value to terminate the EM and the constrained iteratively reweighted least squares (CIRLS) algorithm.
maxit	The maximum number of iterations allowed for the CIRLS algorithm.

Details

The transformation-free linear regression for compositional responses and predictors is implemented. The function to be minized is -\sum_{i=1}^ny_i\log{y_i/(X_iB)}. This is an efficient self implementation of the EM algorithm of Fiksel, Zeger and Datta (2022).

The function tflr.irls() uses the CIRLS algorithm and is faster, but the resulting value is slightly higher in the 4rth of 5th digit. Also, this function is used to obtain initial values for the tflr(), rendering it faster than previous implementations where the scls was used for the initial values.

Value

A list including:

kl	The Kullback-Leibler divergence between the observed and the fitted response compositional data.
be	The beta coefficients.
est	The fitted values of xnew if xnew is not NULL.

Author(s)

Michail Tsagris.

R implementation and documentation: Michail Tsagris [e-mail].

References

Fiksel J., Zeger S. and Datta A. (2022). A transformation-free linear regression for compositional outcomes and predictors. Biometrics, 78(3): 974–987.

Tsagris. M. (2025). Constrained least squares simplicial-simplicial regression. Statistics and Computing, 35(27).

Tsagris M. (2025). Scalable approximation of the transformation-free linear simplicial-simplicial regression via constrained iterative reweighted least squares. https://arxiv.org/pdf/2511.13296

See Also

cv.tflr, scls kl.alfapcr

Examples

library(MASS)
y <- rdiri(214, runif(3, 1, 3))
x <- as.matrix(fgl[, 2:9])
x <- x / rowSums(x)
mod <- tflr(y, x, xnew = x)
mod

Compositional documentation built on Dec. 4, 2025, 9:07 a.m.