FOEH: Frequent Pattern Mining Performance Optimization over Large Transactional Data in Extended Hadoop MapReduce

Abstract

Abstract Frequent item mining is a process wherein we extract or mine frequent itemsets from a given input dataset. Apriori algorithms and FP-growth algorithms are two types of common pattern mining algorithms. Traditional implementations of such frequent item mining algorithms tend to be inefficient when it comes to mining frequent item sets over large transactional datasets, owing to the fact that they consume a greater amount of time in performing the same. In order to tackle the issues mentioned, large transaction datasets using the extended Hadoop MapReduce Framework are considered. On the same, a novel distributed, parallel processing, frequent item mining algorithm is integrated. The analysis of the performance of the implemented algorithm proves that the performance of frequent item analysis in relation to data uploading time, HDFS disk utilization, and data processing time has improved drastically.