Improving the Estimation of Software Development Effort Using the Combination of Cuckoo Search and Particle Swarm Optimization Algorithms

Accurate estimation of required effort for software development has an important role in success of such projects. So far, a lot of research work has been conducted to estimate the effort, but improving the precision of this calculation is still a challenge. In this paper, an approach is proposed based on the metaheuristic algorithms to solve this challenge. The procedure is as follows. First, the Cuckoo Search algorithm is used in order to select the correct software features in estimating effort. Then, the results are further analyzed by Particle Swarm Optimization algorithm. The idea is that the sequential application of these algorithms has led to more accurate search of the problem space and possibility of achieving the global optimum, i.e. the best features is increased. Finally, the selected features are used as the input parameters of the COCOMO II post-architecture model and the effort is estimated. The proposed approach is evaluated on two datasets of COCOMO 81 and COCOMO NASA and in order to its evaluation, two metrics, namely the median magnitude of relative error and the percentage of prediction are used. The results obtained from the experiments of this approach and their comparison to the results of the previous works show that on the COCOMO 81, the value of the median magnitude of relative error decreased by 0.177 and the percentage of prediction, for the three values of 25, 30 and 40 percent, increased by 7.87%, 8.04% and 8.66%, respectively. Furthermore, on the COCOMO NASA, the value of the median magnitude of relative error decreased by 0.151 and the percentage of prediction, for the three values of 25, 30 and 40 percent, increased by 7.55%, 7.98% and 8.11%, respectively.