Implementation of statistical classification methods in stellar catalogs

  • Abstract
    One of the present challenges in Astronomy is to establish efficient statistical methods to perform suitable classification of stellar objects within large databases. In this paper we study the behavior of different statistical methods to perform stellar classification based on several measurements of the star. Specifically, and using JHC stelar catalogue, we show that: (i) there is a relationship between color of the star and the statistical central moments of its spectrum, (ii) we can fit a regression model between star temperature and spectrum shape; (iii) we make an stellar classification based on kernel methods (iv) it is possible to construct an adaptive stelar classification using Kohonen maps.
  • Keywords: Astrostatistics, Classification, Kohonen maps.
  • AMS Subject classifications: 62P35, 62H30.