Enhancing Speed of Map Reduce Classification Algorithms Using Pre-Processing Technique

All Rights Reserved © 2016 IJARCET 2482  Abstract— With exponentially increasing electronic data day by day, Big Data is gaining attention for solving faster access and summarization problems. However, this huge amount of data with heterogeneous formats compelled us to renovate our traditional use of learning algorithms and ponder about new techniques which are challenging and complex. To solve problem of big data, we propose a linguistic fuzzy rule based classification system, which mainly consist of two methods viz. FuzzyReducerMax and FuzzyReducerAve. As name fuzzy suggest vague and uncertain in the similar way it is dealing with uncertainty that is essential to the diversity and authenticity of big data and because of the procedure of linguistic fuzzy rules it is capable to render a recognizable and operational classification model. This process is established on the MapReduce framework, which are very popular and frequently used to handle big data by Hadoop framework. The performance measure is done on these methods by using a Data set of networking attack logs. The result shows its capability to provide accuracy on classification with both the approaches and runtime analysis which shows its speed improvement.