Abstract:
Companies have gathered vast amounts of data about their customers. This paper examines the possibility of using this data to group customers to find novel patterns in the data, which the companies could use to their advantage. The data set to test this hypothesis comes from the social network Facebook, to be clustered using co-clustering which approximates the data using clusters. The algorithm did extract novel information, but it was not feasible
to make this information explicit. Some ideas on how to improve the algorithm on this
deficiency are given.