Chapter 6: Classification Algorithms and Applications

Title: Boost up the performance of naïve Bayesian classifier: using the relevant discriminating attributes

Author(s): Novia Nurain, Rafiul Sabbir, Md. Shahab Uddin, Chowdhury Mofizur Rahman

Address: Department of Computer Science and Engineering, United International University, Dhaka-1209, Bangladesh | Department of Computer Science and Engineering, United International University, Dhaka-1209, Bangladesh | Department of Computer Science and Engineering, United International University, Dhaka-1209, Bangladesh | Department of Computer Science and Engineering, United International University, Dhaka-1209, Bangladesh

Reference: Software, Knowledge, Information Management and Applications (SKIMA 2013) pp. 281 - 291

Abstract/Summary: This paper introduces an incremental approach to boost up the performance of traditional Naïve Bayesian classifier using only the relevant discriminating attributes. Our proposed incremental modified Naïve Bayesian classifier (IMNBC) performs better than the traditional Naïve Bayesian classifier on all the domain, on which we performed the experiments. Besides, the IMNBC also can eliminate more than half of the original attributes. Moreover, the algorithm typically learns faster, even with fewer training examples to reach high accuracy.

Order a copy of this article