The comparative analysis of four kinds of exponential modeling methods

To tackle the situation that the actual observation data are not enough, several methods including unbiased GM (1,1) Model, Ratio Modeling Method, Auto Regressive model and linear-logarithmic modeling Model, are proposed to be used to establish exponential model. Since different modeling has its own characteristics, and the fitting results of each modeling are different, hence it is difficult to evaluate and select these four methods. In this paper, according to the data of different rules and methods of numerical simulation, it is suggested to taking the root mean square error as an indicator. The advantages and disadvantages of fitting results of four methods are compared and conclusions are discussed, which provide some valuable reference to build a suitable exponential model.