Closure to 'GA-Based Support Vector Machine Model for the Prediction of Monthly Reservoir Storage' by Jieqiong Su, Xuan Wang, Yong Liang, and Bin Chen

摘要

The authors wish to thank the discussers for expressing the views that initial conditions will greatly affect the run results with genetic algorithm (GA); the authors agree with the views. They would like to stress the following important points of view to provide a reference for potential researchers: 1. The discussers state that the "computational complexity is the main factor to consider in employing the algorithm to achieve an optimal solution in a random search structure." Actually, the computational complexity is closely related to initial conditions settings for GA. With the same genetic operators, the impacts of not only the initial population's characteristics and quantity but the determination of crossover and mutation rates cannot be ignored during the calculation process. The unreasonable distribution of initial population may cause the concentration in a local region for individuals, which is not conducive to the expansion of the search space and the search of global optimum point. Therefore, it is the first problem to be solved for improving global convergence to ensure the diversity of initial population and the rationality of the individual distribution. In addition, improper selection of the crossover rate and the mutation rate will make the convergence speed towards the optimum solution slower, and even cause the system to be stuck in a local optimum. In sum, reasonable selection of the initial population, the crossover rate, and the mutation rate are helpful to decrease the computational complexity and achieve the convergence with an appropriate rate.