A novel fractional time-delayed grey model with discrete fractal derivative and its applications in predicting enrollments and educational expenditure.

Accurate forecasting of the enrollment scale of higher education and education expenditure is the key to scientific decision-making in education. Nevertheless, education data sets are usually small and affected by uncertainties including economy, education system policies, and so on, which result in difficulty in modeling. To address the tissues, we presented a novel time-delayed grey model based on a fractional-order accumulation time term, abbreviated as FHTDGM. The hyperparameters of the model are optimized using the particle swarm optimization. The experiment's results show that the FHTDGM model can fit and forecast enrollment scale and educational spending data with greater accuracy than a group of other grey models, and the forecast MAPEs are 2.396 and 1.244, respectively. The accurate prediction contributes to constructive suggestions for the decision-making of education. [ABSTRACT FROM AUTHOR]