A New Approach to Improve the Overall Accuracy and the Filter Value Accuracy of the GM(1,1) New-Information and GM(1,1) Metabolic Models

摘要

Grey system theory has many facets, one of which is the so-called GM(1,1) model, used for predicting and forecasting. This paper proposes a novel way of improving the overall relative accuracy of the new-information grey model, and the metabolic grey model, and also by improving the filter value accuracy. By incorporating a weight sequence that is populated by a genetic algorithm to minimize the error of the simulated values. The least square parameters (-a) and b, can then be scaled by the values contained in the weight sequence, until a satisfactory result is obtained. If a high level of accuracy can be attained for the simulation values of the model, and also for the filter value, it will ultimately allow for greater forecasting ability.