P-FHM+: Parallel high utility itemset mining algorithm for big data processing
Krishan Kumar Sethi, Dharavath Ramesh, Damodar Reddy Edla · Procedia Computer Science · 2018
High utility itemset (HUI) mining is emerging as an effective pattern mining technique, which discovers itemsets with their utility more than user defined utility threshold. Most of the HUI mining algorithms discover high utility patterns of all length ( k -itemset), while many applications require the HUIs of certain lengths. FHM+ is a HUI mining algorithm which includes the length constraint and produces HUIs of the user defined lengths. FHM+ is suitable to run small size of data on standalone system. In the era of big data, the standalone system are not efficient to process huge transaction data. To accommodate this, a parallel version of FHM+ is proposed and named as P-FHM+ that process big transaction data in distributed manner on a multi-node cluster. Extensive experiments are conducted with proposed P-FHM+ on multiple real time datasets and observed that P-FHM+ outperforms FHM+ in terms or speed and scalability.