Big Data Analytics Using Multi-Classifier Approach with Rhadoop

Big Data Analytics Using Multi-Classifier Approach with Rhadoop

Big Data

BigData Analytics

Data visualization

Decision Tree

Decision trees

K-Nearest Neighbor

Machine learning algorithms

Multi-Classifier

Naive Bayes

Statistical analysis

Testing

Training

Big Data is the massive amount of data that is generated at such a high speed that is very difficult to analyze with traditional tools. Hadoop provides distributed storage and processing, to extract useful information from such huge data. On the other hand, R is open-source data analysis and programming language that facilitates statistical analysis and data visualization. But R is not scalable, it becomes difficult to process big data using R due to its memory limitations. To utilize data visualization, data transformation capabilities of R on Big Data, in this paper we have integrated R with Hadoop using RHadoop[] package and implemented map reduce form of K-Nearest Neighbor, Naive Bayes and Decision Tree Classifiers in R. In this paper we have also implemented Multi-Classifier to improve the accuracy of classification. Multi-Classifier combines the power of individual classifier to increase the eciency and accuracy of classication. We have used Bayesian combinatorial function and majority voting to combine powers of the above mentioned classifiers. We have found that Multi-Classifier approach gives an improvement in parameters like precision, recall and accuracy.

478-484

P. Hiranandani

E. S. Pilli

N. Chand

C. Ramakrishna

M. Gupta

2018 8th International Conference on Cloud Computing, Data Science & Engineering (Confluence)

2018

conferencePaper

10.1109/CONFLUENCE.2018.8442876