Analysis and comparison of HUPID and FHIM algorithm for dynamic databases

High utility itemset mining provides more useful and realistic results than frequent pattern mining because of its ability to consider statistical correlation and semantic significance among the items. The state of art algorithms designed for mining high utility itemsets always consider the database as static. If they are used for dynamic databases for the same purpose, database is rescanned from the beginning for each update in data. If there is a provision to add new data to the previous analysis results, existing and newly generated data can be efficiently handled and mining can be made more effective with reduced overhead of database scans for dynamic databases. This paper compares performance of HUPID-Tree algorithm over the state of art algorithm FHIM for mining high utility itemsets from dynamic databases and proves that HUPID-Tree algorithm performs better in terms of space and time because it uses incremental mining strategy and local itemset pruning technique.