Mining Approximate Frequent Itemsets over Data Streams Using Window Sliding Techniques

Frequent itemset mining is a core data mining operation and has been extensively studied in a broad range of application. The frequent data stream itemset mining is to find an approximate set of frequent itemsets in transaction with respect to a given support threshold. In this paper, we consider the problem of approximate that frequency counts for space efficient computation over data stream sliding windows. Approximate frequent itemsets mining algorithms use a user-specified error parameter, E, to obtain an extra set of itemsets that are potential to become frequent later. Hence, we developed an algorithm based on the Chernoff bound for finding frequent itemsets over data stream sliding window. We present an improved algorithm MAFIM (a maximal approximate frequent itemsets mining) for frequent itemsets mining based on approximate counting using previous saved maximal frequent itemsets. The proposed algorithm gave a guarantee of the output quality and also a bound on the memory usage.