Hybrid Feature Selection based Classifier for Non Functional Requirements


  • Naina Handa, Dr. Anil Sharma, Dr. Amardeep Gupta




Requirements can be segregated into Functional and Non Functional Requirements. The classification and sub-classification of requirements is a very crucial task. Feature Selection (FS) plays a significant role in classification. The goal of FS is to identify the most important aspects of a problem domain. It is quite beneficial in terms of increasing computing speed and accuracy. Identifying meaningful features from hundreds or even thousands of similar features is a difficult task. In this research, we present a hybrid FS technique for NFRs classification that incorporates two FS methods – filters and wrappers. Candidate features are initially chosen from the original feature set using two filter approaches, and then the intersection of these two sets is refined using more precise wrappers with the Bayesian Belief Model. Both the filters and the wrappers are used in this hybrid technique. The mechanism is investigated using a primary dataset of NFRs. The results of the experiments demonstrate that using a reduced feature set can improve prediction accuracy.




How to Cite

Dr. Amardeep Gupta, N. H. D. A. S. (2022). Hybrid Feature Selection based Classifier for Non Functional Requirements. Mathematical Statistician and Engineering Applications, 71(2), 01–11. https://doi.org/10.17762/msea.v71i2.61