****ing around with a high volume, low quality dataset, using loopy belief propogation to assume missing variables in order to get something out of the data we're actually able to capture, ...