Machine learning autoclassification of candidate variable stars in Python
Identification and classification of variable stars provides valuable data for understanding stellar population composition, structure and evolution. Lightcurve inspection allows detection of intrinsic pulsators, rotationally modulated variables, eclipsing binaries and other exotica. The volume of relevant lightcurve data pending analysis is considerable, so much so that some associated crowdsourced citizen science analysis efforts are projected to extend years. The efficacy of a Python-based machine learning system using multinomial logistic regression, for automating or complementing such efforts, is examined.
