•  
  •  
 

Turkish Journal of Electrical Engineering and Computer Sciences

Author ORCID Identifier

TAPAN PRADHAN: 0009-0004-8295-4023

RADHAMOHAN PATTANAYAK: 0000-0003-3053-4858

Abstract

In recent times, software fault prediction (SFP) has become a crucial task to maintain the quality and reliability of software products in the complex, high-dimensional, and nonlinear data scenario. SFP necessitates a huge volume of historical software metrics data to efficiently identify the hidden patterns, trends, and relations among those metrics. This huge volume of data often contributes to the issue of high-dimensionality and its ramifications in the prediction models, due to the presence of less informative and redundant features in the dataset. Therefore, to tackle the above issues, in this paper a novel wrapper feature selection model employing the starfish optimization algorithm (SFOFS) is proposed. This model identifies the optimal software metrics from the high-dimensional fault dataset by eliminating unnecessary, insignificant, and noisy metrics. The bioinspired metaheuristic algorithms are inherently susceptible to the issue of diversity degradation, leading to premature convergence and suboptimal solutions. Hence, the Levy mutation strategy is incorporated into the SFO algorithm, which improves the diversity of the population by using a heavy-tailed Levy distribution. Further, different existing state-of-the-art models and several baseline models involving widely adopted metaheuristic algorithms are employed to compare and gauge the prediction efficacy of the present SFOFS model. The analysis of the experimental outcome, along with the statistical analysis, showed the SFOFS model outperforming other comparative models for 12 software fault datasets.

DOI

10.55730/1300-0632.4294

Keywords

Feature selection, starfish optimization algorithm, software fault prediction, mutation

First Page

1150

Last Page

1168

Publisher

The Scientific and Technological Research Council of Türkiye (TÜBİTAK)

Creative Commons License

Creative Commons Attribution 4.0 International License
This work is licensed under a Creative Commons Attribution 4.0 International License.

Share

COinS