Preliminary results of using k-Nearest Neighbours Regression to estimate the redshift of radio selected datasets

TL;DR

This study explores using k-Nearest Neighbours Regression to estimate redshifts in large radio survey datasets, demonstrating acceptable accuracy and potential for future astronomical applications.

Contribution

First detailed evaluation of k-Nearest Neighbours Regression for redshift estimation in radio survey data, including tests on various datasets and conditions.

Findings

Achieved ~10% outlier rate in redshift estimation

Method performs acceptably with limited multi-wavelength data

Potential improvement with more uniform training data

Abstract

In the near future, all-sky radio surveys are set to produce catalogues of tens of millions of sources with limited multi-wavelength photometry. Spectroscopic redshifts will only be possible for a small fraction of these new-found sources. In this paper, we provide the first in-depth investigation into the use of k-Nearest Neighbours Regression for the estimation of redshift of these sources. We use the Australia Telescope Large Area Survey radio data, combined with the Spitzer Wide-Area Infrared Extragalactic Survey infra-red, the Dark Energy Survey optical and the Australian Dark Energy Survey spectroscopic survey data. We then reduce the depth of photometry to match what is expected from upcoming Evolutionary Map of the Universe survey, testing against both data sets. To examine the generalisation of our methods, we test one of the sub-fields of Australia Telescope Large Area Survey against the other. We achieve an outlier rate of ~10% across all tests, showing that the k-Nearest Neighbours regression algorithm is an acceptable method of estimating redshift, and would perform better given a sample training set with uniform redshift coverage.