Automated classification of variable stars for ASAS 1-2 data

Abstract

With the advent of surveys generating multi-epoch photometry and the discovery of large numbers of variable stars, the classification of these stars has to be automatic. We have developed such a classification procedure for about 1700 stars from the variable star catalogue of ASAS 1-2 (All Sky Automated Survey, Pojmanski 2000) by selecting the periodic ones and by applying an unsupervised Bayesian classifier using parameters obtained through a Fourier decomposition of the light curve. For irregular light curves we used the period and moments of the magnitude distribution for the classification. In the case of ASAS 1-2, 83% of variable objects are red giants. A general relation between the period and amplitude is found for a large fraction of those stars. The selection led to 302 periodic and 1429 semi-periodic stars which are classified in 6 major groups: eclipsing binaries, "sinusoidal curves", Cepheids, small amplitude red variables, SR and Mira stars. The type classification error level is estimated to be about 7%.