Mining frequent itemsets in data streams based on genetic algorithm

Stream data is a very common data type in big data and in many data streams applications, users tend to pay more attention to the mode information of the data streams. So mining frequent patterns in data streams is a significative work. Meanwhile, finding frequent itemests in a data set with predefined fixed support threshold could be seen as an optimization problem. In this paper, the problem of frequent itemsets mining is derived as a non-linear optimization problem, then genetic algorithm is adopted to solve it. Through the formal and bitmap representation of frequent itemsets, the non-linear optimization problem is transformed to 0-1 programming. A set of experimental results show that unlike typical Apriori algorithm, the complexity of time and memory space grows exponentially as the support decrease, our proposed algorithm has a high time and space efficiency even with a very low support.