Unveiling the stellar halo with TGAS

The detailed study of the Galactic stellar halo may hold the key to unlocking the assembly history of the Milky Way. Here, we present a machine learning model for selecting metal poor stars from the TGAS catalogue using 5 dimensional phase-space information, coupled with optical and near-IR photometry. We characterise the degree of substructure in our halo sample in the Solar neighbourhood by measuring the velocity correlation function.