We apply machine learning techniques in an attempt to predict and classify stellar properties from noisy and sparse time series data. We preprocessed over 94 GB of Kepler light curves from MAST to classify according to ten distinct physical properties using both representation learning and feature engineering approaches. Studies using machine learning in the field have been primarily done on simulated data, making our study one of the first to use real light curve data for machine learning approaches. We tuned our data using previous work with simulated data as a template and achieved mixed results between the two approaches. Representation learning using a Long Short-Term Memory (LSTM) Recurrent Neural Network (RNN) produced no successful predictions, but our work with feature engineering was successful for both classification and regression. In particular, we were able to achieve values for stellar density, stellar radius, and effective temperature with low error (~ 2 - 4%) and good accuracy (~ 75%) for classifying the number of transits for a given star. The results show promise for improvement for both approaches upon using larger datasets with a larger minority class. This work has the potential to provide a foundation for future tools and techniques to aid in the analysis of astrophysical data.

We present H- and K-band spectra for late-type Kepler Objects of Interest (the "Cool KOIs"): low-mass stars with transiting-planet candidates discovered by NASA's Kepler Mission that are listed on the NASA Exoplanet Archive. We acquired spectra of 103 Cool KOIs and used the indices and calibrations of Rojas-Ayala et al. to determine their spectral types, stellar effective temperatures and metallicities, significantly augmenting previously published values. We interpolate our measured effective temperatures and metallicities onto evolutionary isochrones to determine stellar masses, radii, luminosities and distances, assuming the stars have settled onto the main-sequence. As a choice of isochrones, we use a new suite of Dartmouth predictions that reliably include mid-to-late M dwarf stars. We identify five M4V stars: KOI-961 (confirmed as Kepler 42), KOI-2704, KOI-2842, KOI-4290, and the secondary component to visual binary KOI-1725, which we call KOI-1725 B. We also identify a peculiar star, KOI-3497, which has a Na and Ca lines consistent with a dwarf star but CO lines consistent with a giant. Visible-wavelength adaptive optics imaging reveals two objects within a 1 arc second diameter; however, the objects' colors are peculiar. The spectra and properties presented in this paper serve as a resource for prioritizing follow-up observations and planet validation efforts for the Cool KOIs, and are all available for download online using the "data behind the figure" feature.