A High Utility Itemset Mining Algorithm Based on Particle Filter

High utility itemset mining is an interesting research in the field of data mining, which can find more valuable information than frequent mining. Several high-utility approaches have already been proposed; however, they high computational costs and low efficiency. To solve this problem, a algorithm based on particle filter proposed. This approach first initializes population, consists sets. Then, to update sets their weights, novel state transition model suggested. Finally, alleviates degradation problem by resampling. Substantial experiments UCI datasets show that proposed outperforms other previous algorithms terms efficiency, number itemsets, convergence.