An optimized conformable fractional non-homogeneous gray model and its application

Abstract

Developing a robust, accurate forecasting model and improving the prediction abilities of the limited historical data that lacks statistical rules has become a top priority. To address this problem, an improved conformable fractional non-homogeneous gray model, namely CFONGM(1,1,k,c), is proposed. Combining the dynamic background-value and particle swarm optimization algorithm to further improve forecasting ability of the existing gray model. In which matrix perturbation theory is employed to prove that the novel model conforms the principle of new information priority and has a smaller perturbation bound of solution. The two empirical examples of educational funds and new students enrollment of regular institutions of higher education of China are employed to examine the prediction accuracy of the novel model. The results show that the novel model has a better prediction performance compared with other competitive models.