Mining Frequent Patterns from Data streams using Dynamic DP-tree

A DataStream is a real time continuous, ordered sequence of items. It is impossible to control the order in which items arrive, nor it is feasible to locally store a stream in reality. By short it is a rapid flow of continuous ordered data. By these specific characteristics static models and static two pass algorithms are not suitable to data streams. Data stream mining have following three challenges one every item is examined only once. Second the storage space should control even there is a large amount of data, third the mining results have to be produced as early as possible. In this paper we propose a novel method to mine the frequent items over data streams by dividing data as no of windows and mine frequent item sets over window using a very compact data structure DP-Tree and placing the every DP-Tree safely in disk space so that we can retrieve the tree structure for pruning as and when we require. More over we propose methods to dynamically construct and update the DP-Tree