Grey System Theory And The Comparison Of Related Models

With the rapid development of society and economy, The reasonable prediction to the future trend become to be not only the primary pursuit of all walks of life, but also the fundamentality of various industries on the social incentive competition stage, so all kinds of forecasting methods are applied to many fields.Multivariate linear regression analysis model is one of the many forecasting method, it can be used to format the connection between one or more independent prediction variables and a (continuous valued) response variables modeling. In data environment, prediction variables describe the interested attributes of tuple and a response variable is to be predicted. At present, they are widely used in natural and social science.To ensure that model has good explanation power and forecasting effect, in establishing a multiple linear regression model, we should do the following: First of all, dependent variables must have significant effect on independent variables and the linear correlation between them is closely. The linear correlation can not is a form and must be true; Secondly, The related degree between independent should not be higher than the correlation between the independent variable and dependent variable, and dependent variables should be relatively complete statistics. Therefore, multivariate linear regression model has its disadvantages. Multivariate linear regression analysis does not consider the time factor for the effects of prediction, so it has slow perception for dependent variable and it is more sensitive to the pathological data in the sample. If appearing pathological data, it will affect fitting effect, which causes inaccurate prediction. But in practice, we should not only consider the dynamic changes of the dependent variable, but also reduce the effect of pathological data to the influence of fitting effect. This need to look for new and more excellent models. Based on the outstanding characteristic of the time sequence of the gray system theory, so it can track the dynamic changes of the dependent variable and avoid the effect of the pathological data to fitting effect. Grey dynamic model models based on gray generating function and the differential fitting as the core, the thought is to converse the time sequence into differ- ential equation as to as establish a abstract dynamic model of the develop- ment changes. Based on the thought of two models, the article establish a new model, namely the ash multivariate linear regression analysis model. A lot of experiments show that the new model can make more and true accurate prediction results and the predictive results can become theoretical basis for the decision making of each field.