Application Of Random Forest In Photometric Redshift

Redshift is of great significance to the formation and evolution of the cosmic large-scale structure(LSS).With the development of large sky survey project,it is particularly important to accurately estimate the photometric redshift and fully understand the uncertainties caused by systematic error.In this paper,the machine learning algorithm of random forest is used to estimate the photometric redshift of China Space Station Optical Survey.The training set data used for the test is simulated from the COSMOS catalog.Through this algorithm,the redshift probability distribution function(PDF)of each galaxy can be obtained.In addition,after further digging into the algorithm,we can get the importances of input features,the confidence of predicting redshift and other information.Comparing the results with the template fitting method,it is found that although the results do not depend on the completeness of the template,they are still limited by the distribution of the training set.