Abstract
In this paper, a GM(1, 1) polymorphic management prediction algorithm based on a dormant system is proposed, which first obtains the optimal data quantity m through simulation, and then the data used is forecasted only once, and the input data will be updated to the latest data immediately after the prediction is completed, and then the parameter estimates will be recalculated for the next prediction. The simulation results show that the MRE value of the optimized GM (1, 1) is 0.090, which has much better MRE values for GM (1, 1) algorithms that are not optimized. At the same time, other commonly used algorithms in the industry are compared horizontally. In contrast, the optimized GM (1, 1) algorithm responds quickly, has high prediction accuracy, short prediction time, and consumes fewer resources.

This publication has 3 references indexed in Scilit: