Identification of feature selection techniques for software defect prediction by using BCF-WASPAS methodology based on Einstein operators

Abstract

Listen

Translate article

PurposeThis research focuses on a very important research question of determining the appropriate feature selection methods for software defect prediction. The study is centered on the creation of a new method that would enable the identification of both positive and negative selection criteria and the handling of ambiguous information in the decision-making process.Design/methodology/approachTo do so, we develop an improved method by extending the WASPAS assessment in the context of bipolar complex fuzzy sets, which leads to the bipolar complex fuzzy WASPAS method. The approach also uses Einstein operators to increase the accuracy of aggregation and manage complicated decision-making parameters. The methodology is designed for the processing of multi-criteria decision-making problems where criteria have positive and negative polarities as well as other ambiguous information.FindingsIt is also shown that the proposed methodology outperforms the traditional weighted sum or product models when assessing feature selection methods. The incorporation of bipolar complex fuzzy sets with WASPAS improves the assessment of selection criteria by taking into account both positive and negative aspects of the criteria, which contributes to more accurate feature selection for software defect prediction. We investigate a case study related to the identification of feature selection techniques for software defect prediction by using the bipolar complex fuzzy WASPAS methodology. We compare the proposed methodology with certain prevailing ones to reveal the supremacy and the requirements of the proposed theory.Originality/valueThis research offers the first integrated framework for handling bipolarity and uncertainty in feature selection for software defect prediction. The combination of Einstein operators with bipolar complex fuzzy sets improves the DM process, which will be useful for software engineers and help them select the best feature selection techniques. This work also helps to enhance the overall performance of software defect prediction systems.