The Journal of Grey System ›› 2025, Vol. 37 ›› Issue (6): 53-65.

Previous Articles     Next Articles

A Time Delayed Discrete Grey Power Model with Dynamic Background Value and Its Application in Forecasting Master's Enrollment Scale#br#

  

  1. 1. School of Computer, Jiangsu University of Science and Technology, Zhenjiang, Jiangsu, 212100, P.R. China
    2. School of Business, East China University of Science and Technology, Shanghai, 200237, P.R. China.
    3. School of Management Science and Engineering, Chongqing Technology and Business University, Chongqing, 400067, P.R. China.
  • Online:2025-12-01 Published:2025-12-29
  • Supported by:
    The authors appreciate the financial support provided by the Scientific Research Startup Project for High-level Talents of
    Chongqing Technology and Business University (2555012) and the Planning Project of Humanities and Social Sciences Research of Chongqing Municipal Education Commission (23SKGH164).

Abstract: In recent years, the "postgraduate entrance exam fever" has become a hot topic in public discourse. The contradiction between the supply of talent cultivation in higher education institutions and societal demands has become increasingly prominent. Therefore, it is crucial to guide graduates in rationally understanding the future trends of postgraduate entrance exams, and the quantity of master's enrollment is a key factor influencing whether undergraduates choose to apply. Based on the three-parameter discrete grey forecast model, this paper introduces dynamic background value and a time delayed power term to construct a time delayed discrete grey power model (TDDBGM(1,1,λ,tα). With the goal of minimizing the MAPE of total test, the PSO (Particle Swarm Optimization) algorithm is employed to globally optimize the parameters λ and α. Subsequently, this model is used to simulate and forecast the quantity of master's enrollment under the background of higher education popularization. The results demonstrate superior performance with train set error (3.28%), test set error (1.39%), and total set error (2.94%), all of which are significantly lower than those of the TDDGM(1,1,tα) model, DBGM(1,1) model, GM(1,1,t2) model, ARIMA model and SVR model. Based on the results, this
study proposes relevant policy recommendations.

Key words: Grey forecast model, Dynamic background value, Time delayed power term, PSO, Master's enrollment scale