Variable Selection Using Nearest Neighbor Gaussian Processes

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Posch, Konstantin, Arbeiter, Maximilian, Truden, Christian, Pleschberger, Martin, Pilz, Juergen
Format: Preprint
Published: 2021
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913515983339520
author Posch, Konstantin
Arbeiter, Maximilian
Truden, Christian
Pleschberger, Martin
Pilz, Juergen
author_facet Posch, Konstantin
Arbeiter, Maximilian
Truden, Christian
Pleschberger, Martin
Pilz, Juergen
contents We introduce a novel Bayesian approach for variable selection using Gaussian process regression, which is crucial for enhancing interpretability and model regularization. Our method employs nearest neighbor Gaussian processes, serving as scalable approximations of classical Gaussian processes. Variable selection is achieved by conditioning the process mean and covariance function on a random set that represents the indices of contributing variables. A priori beliefs regarding this set control the variable selection, while reference priors are assigned to the remaining model parameters, ensuring numerical robustness in the process covariance matrix. We propose a Metropolis-Within-Gibbs algorithm for model inference. Evaluation using simulated data, a computer experiment approximation, and two real-world data sets demonstrate the effectiveness of our approach.
format Preprint
id arxiv_https___arxiv_org_abs_2103_14315
institution arXiv
publishDate 2021
record_format arxiv
spellingShingle Variable Selection Using Nearest Neighbor Gaussian Processes
Posch, Konstantin
Arbeiter, Maximilian
Truden, Christian
Pleschberger, Martin
Pilz, Juergen
Computation
We introduce a novel Bayesian approach for variable selection using Gaussian process regression, which is crucial for enhancing interpretability and model regularization. Our method employs nearest neighbor Gaussian processes, serving as scalable approximations of classical Gaussian processes. Variable selection is achieved by conditioning the process mean and covariance function on a random set that represents the indices of contributing variables. A priori beliefs regarding this set control the variable selection, while reference priors are assigned to the remaining model parameters, ensuring numerical robustness in the process covariance matrix. We propose a Metropolis-Within-Gibbs algorithm for model inference. Evaluation using simulated data, a computer experiment approximation, and two real-world data sets demonstrate the effectiveness of our approach.
title Variable Selection Using Nearest Neighbor Gaussian Processes
topic Computation
url https://arxiv.org/abs/2103.14315