Graduate enrollment prediction by an Error Back Propagation algorithm based on the Multi-Experiential Particle Swarm Optimization

The graduate enrollment is influenced by the current national policy, the social needs, and the social economic status and so on. The change of the enrollment number shows the nonlinearity and the complexity. In order to have better grasp of the enrollment scale and to realize the rational allocation of educational resources, we propose a Multi-Experiential Particle Swarm Optimization (MEPSO) algorithm. The algorithm is combined with the Error Back Propagation (BP) algorithm to establish a new neural network that is called the MEPSO-BP neural network. Then we present the simulation numerical studies based on several typical algorithms. The results show the MEPSO-BP algorithm improves the convergence speed and the predictive accuracy, and it can be regarded as a new method for the graduate enrollment prediction.