| At present, there are many methods of prediction, but most methods are based on a large number of samples to implement or acquired higher accuracy.But in some circumstances we can not collect integrated sample data, or due to the nature of some study the sample we can select is very limited. Grey forecast aims at poor information, small sample size prediction of the research object. Markov model combining Grey theory can avoid considering other factors'influence, and also can predict research object'state.This paper mainly starts with the basic theories of grey model and markov chain. To traditional GM (1,1) model being a gray index model with deviation, the problem not high of the model precision exists. This paper proposes the residual GM (1,1) and metabolism GM (1,1), combining Markov shift probability reflect all random influence degree and inherent regular advantage that shift between the every state of factor. Fully utilize information of initial data. Solve getting gray to predict model predict low problem of precision to random fluctuation than heavy several. The combination of the two can avoid considering other many kinds of influence factors and predict the change of period in future. It holds greater science and practicability.It is test that predict validity of model, this text go on gray model to the whole country and total import and export value of Association of South-east Asian Nations, the metabolism GM (1,1) and gray Markov predict. Use gray modeling to Yunnan's total import and export value of Association of South-east Asian Nations and Guangxi's total import and export value of Association of South-east Asian Nations compared with incomplete difference GM (1,1) and getting gray Markov predict model in total import and export value of Association of South-east Asian Nations. It Proves getting gray Markov prediction accuracy of model higher than gray to predict the model has higher actual using value. |
