Incremental mining of periodic high utility itemset using artificial bee colony and pruning strategies for retail industry and clinical databases

Abstract

The extensive growth of computer and internet has produced an enormous amount of data These data contain very useful information hidden inside To extract the hidden information from the huge databases data mining algorithms have been evolved Frequent pattern mining is an important area in data mining It discovers the frequently occurring itemset in the database based on the user specified minimum threshold value Association rule mining is another important kind of mining which finds the association that exists between the items Association rule mining and frequent pattern mining algorithms consider the number of occurrences of an item in the data set In reality each item will have different weigh for the attributes such cost weight temperature etc Mining the itemset considering the attributes is high utility itemset mining The utility of the itemset is calculated based on the internal and external utility If an itemset satisfies the user given threshold it is regarded as high utility itemset High utility itemset is applied in various fields including market basket analysis weather prediction business decision making etc The existing algorithms to mine the high utility itemset use a recursive procedure to explore the search space This is because the downward closure property is not satisfied in high utility itemset mining Even though many algorithms have been proposed to effectively mine the high utility itemset the search space cannot be reduced to a greater extent newline