Linear Sufficient Dimension Reduction

A package for some SDR methods.

The package linearsdr contains some implementations for simple sufficient dimension reduction methods, including Sliced Inverse Regression (SIR), Sliced Average Variance Estimator (SAVE), Directional Regression (DR) and Outer Product of Gradients (OPG). The package also contains implementation of the Outer Product of Canonical Gradients (OPCG) and Minimum Average Deviance Estimation (MADE) from the paper

• Quach, H. and Li, B. “On Forward Sufficient Dimension Reduction for Categorical and Ordinal Responses”. Electron. J. Statist. 17 (1) 980 - 1006, 2023. https://doi.org/10.1214/23-EJS2122. Link.

The package, along with some examples, can be found here

• Contains R Code for the Outer Prodcut of Canonical Gradients (OPCG) and Minimum Average Deviance Estimation (MADE) methods developed in

• Contains R Code for other linear SDR methods: OPG, MAVE, SIR, SAVE, and DR

• Provides a Tikhonov regularization option for SIR, SAVE and DR to naively handle rank-deficient covariance matrices.