Objectives: Feature Selection is a important technique to reduce the quantity of features in various application domains where the data set includes thousands of features in it. The main objective of this study is to choose BFPA for feature selection to obtain better ﬁtness function. Methods/Statistical Analysis: Improved Binary Particle Swarm Optimization (iBPSO) approach is used for selecting the subset from the dataset and providing the better ﬁtness values. Here, iBPSO used to obtain the feature subset and its performance is compared with BFPA. Na¨ıve Bayes classiﬁer is used to improve the classiﬁcation accuracy. Findings:The experimental results shows that BFPA overall accuracy is improved 0.7% in using NB and 0.4% using k-NN as compared to iBPSO. Application/Improvements: To reduce the complexity and increase the accuracy the BFPA is used. Performance analysis shows that BFPA outperforms the iBPSO.
Digital Object Identifier (DOI)
Rajamohana, SP.; Umamaheswari, K.; and Abirami, B.
"Performance Analysis of iBPSO and BFPA Based Feature Selection Techniques for Improving Classiﬁcation Accuracy in Review Spam Detection,"
Applied Mathematics & Information Sciences: Vol. 11
, Article 22.
Available at: https://dc.naturalspublishing.com/amis/vol11/iss4/22