Data Processing Technology Of Grey Forecast Model And Its Application Research

Grey system theory is based on small sample and poor information uncertain systems with unknown information and known part of information.GM model is the basic model of grey prediction theory,and it is widely used.It is based on the exponential law of the original data after accumulative processing.However,the traditional forecasting models are suitable for low growth and near-exponential series,and most sequences are neither near-exponential nor low-growth sequences in reality.Therefore,in order to improve the accuracy of modeling,the original sequence should be analyzed and processed before establishing the grey prediction model.This paper will focus on the first step of grey prediction technology--data preprocessing stage.The main contents are as follows:(1)Based on the data preprocessing technology,that is,from two aspects of function transformation and buffer operator,new data processing methods are found and modeling accuracy is improved.(2)Features and parameter range of the function transformation with parameter in proving the modeling accuracy.Firstly,this paper research on the parameter range of the transformation with parameter which can reduce class ratio dispersion.However,in view of the fact that the forecast values are not required,these values must be reduced,and in the process of reduction,some transformations further improve the modeling accuracy,while some enlarge the relative error,transformations further improve the modeling accuracy,while some enlarge the relative error.Therefore,the paper then research on the function feature that does not enlarge relative error during the reversion process.Finally,an example is given to prove the operability of this paper and the reasonableness and reliability of conclusion.(3)A Class of Weakening Buffer Operators for Reducing the Class Ratio Dispersion.Compared with the function transformation,the superiority of the buffer operator is that it does not need inverse transformation,so a new weakening buffer operator is proposed in this paper.First,the appropriate function transformation is used to reduce class ratio dispersion,and the necessary paving is made for the buffer operator.Then the translation constant is determined based on the value of the last term of the sequence,so that the sequence transformation satisfies the fixed point axiom of the buffer operator,then a weakening buffer sequence is obtained.Thus effectively avoiding the limitation of function transformation reduction and enlarging the accuracy of modeling.Thus effectively avoiding the limitation that the function transformationmay enlarge relative error during the inverse transformation process,and improving the modeling accuracy.