UPSILoN

[ UPSILoN Logo ]

UPSILoN stands for "AUtomated Classification of Periodic Variable Stars Using MachIne Learning" (Kim & Bailer-Jones 2015).

As the name tells, it classifies periodic variable using machine learning and variability. It also classifies light curves from any astronomical time-series surveys in optical bands regardless of their survey-specific characteristics (e.g. sampling rates, colors, etc).

It mainly consists of two parts as:

1. extracting variability features from light curve(s)
   • derives multiple features from several literatures (e.g. Kim+ 2011, 2012, 2014)
2. classifying the light curves
   • using Random Forest (RF)

The classification model is trained/tested using light curves from many different surveys such as EROS-2, OGLE, MACHO, LINEAR, ASAS, and Hipparcos.

The package is developed using Python + scikit-learn. If you have any questions, please contact me. If you use UPSILoN in scientific publication, we would appreciate citations to Kim & Bailer-Jones 2015.

GitHub: https://github.com/dwkim78/upsilon

PyPi: https://pypi.python.org/pypi/upsilon

Keywords: astronomy - periodic variables - light curves - variability features - time-series survey - machine learning - classification