Tuesday

The importance of accurate photometric redshifts will increase in the coming years, as deeper and wider imaging surveys (such as DES, LSST and EUCLID) become increasingly important for constraining various cosmological parameters and studying galaxy evolution. A promising new method that has proven to provide efficient, accurate photometric redshift estimations coupled with accurate variance predictions is the machine learning method Gaussian Processes for photometric redshift estimation (GPz) (Almosallam et al. (2016)). We investigate a number of methods for improving the photo-z estimations obtained from this algorithm. We use spectroscopy from the Galaxy and Mass Assembly Data Release 2 along with corresponding Sloan Digital Sky Survey visible (ugriz) photometry and the UKIRT Infrared Deep Sky Survey Large Area Survey near-IR (YJHK) photometry and find that using additional input parameters (in the form of near-IR filters and angular size data) for the training, validation and testing of the GPz algorithm significantly improves the accuracy of the estimations obtained. In addition, we explore a post-processing method of shifting the probability distributions of the estimated redshifts and find that it unambiguously provides an increase in accuracy of the estimations. Finally, we investigate the effects of using more precise photometry obtained from the Hyper Suprime-Cam Subaru Strategic Program and find that it produces significant improvements in accuracy, similar to the effect of including additional features.