Identification and classification of interesting variable stars in the MACHO database

Abstract

The MACHO database is an astronomical database of the intensities of about 20 million stars, recorded approximately every night for several years. About one percent of these stars are classified as ``variable stars''. These variable stars are generally roughly periodic and can have periods ranging from less than one day to hundreds or thousands of days. We investigate the application of computationally simple features in order to classify these stars. We present a methodology for extracting potentially interesting stars based on their location in feature-space, and methods for using human interaction to group these interesting stars.