This was a project I worked on to analize facility access right data. The data contained categories such as card holders (users), cared reader device ids, card reader groups, card holder deparments, card holder job title, and access groups. A card reader group is a collection of card readers. An access group is a collection of card readers and card reader groups. The objective was to find the best organization of card readers to define an access group that would minimize the number of overlap between indivual card holders. Bernoulli mixture models was tested to see how well it would cluster the data in an unsupervised approach. The bernoulli mixture model was used from the BayesPy package. There was also preprocessing of the data to transform that data into a binary matrix. There are also graphs to visualize the data.