Author : Ms.Swarupa Kulkarni and Prof.Priyanka Kedar

In the existing system Conventional cluster ensemble approaches have several limitations. The must-connect limitation implies that two element vectors ought to be relegated to a similar group, while they can’t interface imperatives implies that the two component vectors can’t be doled out to a similar bunch. A large portion of the group troupe strategies can’t accomplish acceptable outcomes on high dimensional datasets. Conventional example mining calculations are not reasonable for really huge information, displaying two fundamental difficulties to be settled: computational multifaceted nature and primary memory prerequisites. A progression of calculations dependent on the MapReduce system and the Hadoop open-source usage have been proposed here. All the proposed models depend on the notable Apriori calculation and the MapReduce system. The proposed calculations are isolated into three fundamental gatherings. Two calculations Apriori MapReduce (AprioriMR) and iterative AprioriMR (IAprioriMR) are appropriately intended to separate examples in huge datasets. These calculations remove any current thing set in information in any case their recurrence. Pruning the hunt space by methods for the antimonotone property. Two extra calculations space pruning AprioriMR (SPAprioriMR) and top AprioriMR (TopAprioriMR) are proposed with the point of finding any incessant example accessible in information. Maximal successive examples. A last calculation maximal AprioriMR (MaxAprioriMR) is additionally proposed for mining dense portrayals of successive examples, i.e., visit designs with no continuous supersets.

Keywords: Big Data, Hadoop, Data Mining