A kind of new data processing method suited to the GM (1, 1) model

In this paper, we analyze the data transformation given in the literature [8] and the methods to evaluate its parameter proposed in the literature [9], and point out it can´t be suitable to the sequence whose class ratio sequence is unstable and has the increasing or decreasing trend obviously. According to the variation trend of the class ratio sequence of the original sequence, this paper proposes a kind of new data processing method, taking into account the class ratio monotone decreasing or increasing trend. The parameters in the new data transformation are evaluated by the least deviation of class ratio. Examples show that the new method is feasible and efficient.