Title: A wrapper-based feature selection approach using osprey optimisation for software fault detection

Authors: Pradeep Kumar Rath; Soumili Ghosh; Mahendra Kumar Gourisaria; Susmita Mahato; Himansu Das

Addresses: School of Computer Science and Engineering, NIST Institute of Science and Technology Berhampur, Berhampur, India ' School of Computer Engineering, KIIT Deemed to be University, Bhubaneswar, India ' School of Computer Engineering, KIIT Deemed to be University, Bhubaneswar, India ' School of Computer Science and Engineering, NIST Institute of Science and Technology Berhampur, Berhampur, India ' School of Computer Engineering, KIIT Deemed to be University, Bhubaneswar, India

Abstract: Advancements in machine learning, detecting faults in software development cycles alongside determining the level of correctness of said predictions, have improved significantly. This paper employed osprey optimisation in feature selection approach to improve the predictive capabilities of the models. In this paper, the proposed approach takes the feature generation process through two stages, exploration and exploitation phases. The propagation of weight updation through two stages makes sure that all relevant characteristics of the dataset are considered when generating the optimal subset of features. A total of four models were used, namely, K-nearest neighbours (KNN) classifier, naïve Bayes (NB) classifier, decision tree (DT) classifier and quadratic discriminant analysis (QDA), which is a generative model derived as a special case of the NB classifier. The results acquired from the employment were then compared with four other feature selection algorithms, namely, particle swarm optimisation, genetic algorithm, differential evolution and ant colony optimisation. The integration of osprey optimisation algorithm into the prediction process yielded the best results as compared to the counterpart for each model. The statistical analysis of the proposed FSOOA model has been validated using Friedman and Holm procedure for the statistical significance of the results.

Keywords: software fault prediction; feature selection; optimisation techniques; classification; soft computing.

DOI: 10.1504/IJES.2025.144929

International Journal of Embedded Systems, 2025 Vol.18 No.1, pp.1 - 19

Received: 11 Mar 2024
Accepted: 13 Jun 2024
Published online: 11 Mar 2025 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article