Solving Software Project Scheduling Challenges Using a Search-Based Approach

Efficient scheduling is a cornerstone of successful software project management, directly influencing delivery timelines, cost control, and resource utilization. However, inadequate scheduling methodologies remain a critical factor behind project delays and budget overruns, particularly in contexts requiring rapid deployment such as the COVID-19 pandemic. This study addresses the Software Project Scheduling Problem (SPSP) through a search-based approach leveraging metaheuristic optimization techniques. Specifically, we investigate the effectiveness of Genetic Algorithms (GA), Tabu Search (TS), and their hybridization, in comparison with alternative metaheuristics such as the Firefly and Dragonfly algorithms. The International Software Benchmarking Standards Group (ISBSG) dataset, comprising over 2,000 global software projects, is employed as the empirical basis for validation. The research design encompasses a comprehensive literature review, formulation of research questions, and systematic application of GA and GA–TS hybrid models. Experimental evaluation reveals that the hybrid approach achieves substantial improvements over standalone GA, with mean fitness values increasing by approximately 40% across 100 iterations (from 0.948 to 1.887 in the final 10 iterations). Furthermore, the hybrid model reduced convergence time by nearly 30%, enhanced resource allocation accuracy, improved project duration estimates by 25%, and lowered projected costs by 20%. These results demonstrate that the GA–TS hybrid consistently provides more robust, efficient, and reliable scheduling solutions. The findings contribute to both theory and practice by validating the superiority of hybrid metaheuristics in addressing the inherent complexity and nonlinearity of SPSP. They further highlight the potential of search-based techniques to improve real-world project management outcomes in dynamic and resource-constrained environments. Future research should extend this work by integrating additional metaheuristic combinations, incorporating uncertainty modeling, and testing across diverse datasets to generalize applicability and strengthen practical adoption.