Performance Analysis Of Various Data Classification Technique Based On Mapreduce Framework

With the continuous advancement of internet technologies, the importance of efficient data classification has grown significantly. In this research, the performance of selected classification techniques has been analyzed using both a traditional execution framework and a distributed computing framework based on Hadoop MapReduce. Three widely used algorithms—Naive Bayes, Decision Tree, and Linear Support Vector Machine (SVM)—are implemented and tested under varying mapper sizes. Their performance is evaluated using key metrics such as accuracy, precision, recall, memory usage, and computation time. The experimental results are presented through comparison tables, Bar Charts and a Heatmap to highlight the differences in performance across the two frameworks. The findings indicate that the MapReduce-based approach significantly improves classification performance in terms of accuracy and recall, particularly for Decision Tree and Linear SVM, which achieved accuracy rates above 93%, while Naive Bayes also showed noticeable improvement compared to its traditional counterpart. However, these enhancements come at the cost of increased memory consumption—rising from under 1 GB in traditional settings to as high as 5.5 GB in distributed execution—as well as a broader range in computation time depending on the mapper size. This analysis aims to determine how the underlying computational framework influences the efficiency of classification algorithms and to identify which technique performs best under different execution environments.