Hadoop Map Reduce Techniques: Simplified Data Processing on Large Clusters with Data Mining

Abstract

Data mining applications have become outdated and outmoded in recent years. The use of incremental processing to refresh mining results is a promising method. It makes use of previously saved states to save time and energy on re-computation. In this research, we offer a novel increment processing addition to the Map Reduce, the most extensively used methodology for mining the big data by using the Naive Bayes, the J48, and the Random Forest algorithms. Map reduction is a programming model for simultaneous processing and generation of massive amounts of data. We examine Map Reduce employing Naive Bayes, J48, and Random Forest algorithms with a variety of processing features for efficient mining that also saves energy. The Naive Bayes algorithm generates more energy and fewer maps. Priority-based scheduling is a task that allocates schedules based on the jobs’ requirements and utilization. As a result of decreasing the maps, the system’s workload is reduced, and energy efficiency is improved. The experimental comparison of the several algorithm techniques (Naive Bayes, J48, and Random Forest) have applied in this article and found that the Random forest is performed better than remaining two algorithms i.e. 92%.