Mining discriminative itemsets in data streams using different window models

Big data availability in areas such as social networks, online marketing systems and stock markets is a good source for knowledge discovery. This thesis studies how discriminative itemsets can be discovered in the data streams made of transactions out of user profiles. Discriminative itemsets are frequent in one data stream with much higher frequencies than same itemsets in other data streams in the application domain. This research uses heuristics to manage the large and complex datasets by decreasing the number of candidate patterns. This gives researchers a better understanding of pattern mining in multiple data streams