Rapid classification and coarse characterisation of cool stars in K2.
Guy Davies (University of Birmingham), Mikkel N Lund (University of Birmingham), Yvonne Elsworth (University of Birmingham), Bill Chaplin (University of Birmingham)

K2 is providing high-precision photometry on many, many stars. In addition to primary targets, many additional stars are observed in postage stamps and super stamps. Pipelines like K2P2 can easily extract light curves for these ‘bonus’ stars but provide little in the way of classification. We have developed a machine learning scheme based on easily extractable power-spectrum features to classify cool stars and provide a coarse characterisation. We have trained the scheme on both Kepler and K2 data sets. We find that the classification scheme outputs a metric that provides a good estimate of the probability that we will detect solar-like modes of oscillation. Furthermore, using the same power-spectrum features we have trained a supervised regression scheme to provide coarse estimates of global oscillation properties.