Designing and Reporting on Computational Experiments of Multi-objective Component Selection Algorithm

One important issue of component-based software development is the minimization of the development cost and the maximization of the system reliability while satisfying functional requirements. There are numerous publications on this issue based on metaheuristic techniques, but there are two deficiencies: too tough to evaluate the performance of algorithms and fix parameters in the real-world application. To address this problem, a three phased algorithm is proposed by the Wu et al. [International Journal of Information Technology & Decision Making 5 (2011) 811–841]. This paper describes computational experience in solving the problems using the metaheuristics and the proposed algorithm. The results indicate the efficiency of the proposed algorithms in terms of overall nondominated vector generation, well converged set of solutions, and diversity of solutions. Computational results and simulation analysis further assist a decision maker to fix optimal parameters of metaheuristics including the number of iteration, crossover rate, and mutation rate, and explore hints in using metaheuristics for the problem.