An improved approach of FP-Growth tree for Frequent Itemset Mining using Partition Projection and Parallel Projection Techniques

Abstract

Listen

Translate article

In Data mining, it is about analyzing data; about extracting information out of data. It is a very actual as well as interesting issue having more and more data stored in database. The most important usage: customer behavior in market purchasing, shopping cart processed information provide, management of campaign , customer relationship management, mining about web usage called web mining, mining of text. In the current age of science we developed such technology by using it each type of data related to anything such like person, place, shop, or any organization can be stored. By analysis it is found that FP-growth is efficient in terms of tree construction as compared to Apriori and Tree Projection. Tree Projection is faster and more scalable than Apriori. The parallel projection technique is proved to be more scalable than partition projection as partition projection saves memory space as it works well for the dataset which is dispersed, if the FP-growth tree algorithm and Tree Projection are compared on the basis of benefits it holds on, Apriori does not result to be convenient enough. The pros of FP-growth as compared to Apriori concludes to be transparent as the datasets which it contains has an enormous number of combinations of short-narrative frequent patterns. FP-growth tree implemented along with projection techniques i.e. Partition projection technique constructed to reduce execution time for constructing FP-Growth tree has to be carried out.