A Study to Increase the Performance of FP-Growth Method Using Dimension Tree Technique on Huge Volumes of Data

The dominant ARM algorithms viz. Apriori and FP-growth, needs huge resources when the input data is huge. First algorithm is efficient, but it generates intermediate conditional FP trees. The basic idea of our research is to enhance the rendition of FP-growth technique. Therefore, four techniques viz. modified FP-growth algorithm using maximal patters and array strategy, modified FP-growth technique with map reduce step, modified FP-tree using pre-large items and dimension tree usage to efficiently erect the FP-tree were studied here. For all the strategies, different support levels are used to test its efficiency. From the experimental results it is clear that FP-tree constructed using dimension tree is highly compact, needs less scan over the database and resourcefully generates FP-growth tree. It is concluded that usage of dimension tree is efficient than traditional algorithms.