Mining of High Utility Itemsets with Negative Utility values for Incremental Datasets

Recent advances in knowledge discovery have led to the design of efficient methods of extracting meaningful patterns from datasets. Most of the algorithms are designed under the assumption that datasets are inherently static in nature and that all items hold only positive utility value. However, for industrial applications some of the items can have negative utility values to indicate discounts. Also, these databases are likely to evolve with time as new data is inserted. Traditional mining algorithms incur high execution times and demand large memory requirements to process new data by running the algorithms in a batch manner. Therefore, in this study the problem of mining itemsets with high utility, from incremental datasets is addressed, where the items can have negative utility values. Further, a detailed experimental analysis are undertaken to validate the efficacy of the proposed algorithm.