Object-Oriented and Data Flow based Software Defect Prediction using Hybrid Salp Swarm Algorithm and Genetic Algorithm

In recent years, the development of software applications across diverse domains like research, healthcare, satellite operations, and mobile platforms has underscored the necessity for efficient testing methodologies. By integrating data flow principles within object-oriented frameworks, software testing is enhanced to identify potential errors in the code, such as uninitialized variables or inappropriate control flow. However, existing Artificial Neural Network based Input Gain Measurement Approximation (ANNIGMA) has proven effective in object-oriented and data flow testing, but its reliance on mathematical constraints limits its scalability in highly dynamic or large systems. To overcome this limitation, this research proposes a hybrid Salp Swarm Algorithm (SSA) and Genetic Algorithm (SSA-GA) framework, which captures object-oriented structures within software codebases. Initially, a dataset of source code is collected from the PROMISE repository, consisting of a list of classes characterized by code features such as defect labels and lines of code. Following data collection, the SSA algorithm is employed to generate test suites and perform test case selection by efficiently searching the optimal set of test cases for both Object-Oriented Testing (OOT) and Data Flow Testing (DFT). Subsequently, the GA is applied for test suite verification by integrating adaptive mutation and crossover, ensuring faster convergence and reduced redundancy. Experimental evaluation demonstrates that the proposed SSA-GA framework outperforms existing approaches in terms of accuracy (95.2%) and Area Under the Curve (AUC) (0.93) when compared with the ANNIGMA framework.