Predicting 3D-HST redshift with decision trees
This entry explores how to predict galaxy redshift based on their fluxes and colors observed by the Hubble Space Telescope (HST). The 3D-HST team compiled a large survey of several fields in both imaging and spectroscopy and used detailed photometric modeling to infer the redshifts of detected sources. The aim of this entry is to use simple clustering analysis to predict source redshifts, and compare with the published results. Although we find that the simple decision tree model does not perform as well as the published model, the 3D-HST HLSP provides a useful dataset for exploring model parameters.
Data: The 3D-HST HLSP
Notebook: Predicting 3D-HST galaxy redshift with decision trees
Tags: decision trees, 1-d data, regression, cross-validation