| Since the mid1980’s, Multilevel Statistical Model has been widely used in sociology, pedagogy, population, psychology, economics, medical,public health research and other social scientific fields. Its importance is increasing day by day. Especially from the perspective of the development of contemporary, multilevel statistical model is better and can offer more credible results than the traditional regression model, because it is more close to the practice and shows better functionalities. This paper reviews the basic theory of multilevel statistical model and its current research at home and abroad. And based on this, it expounds the modeling ideas, and gives all sorts of model frame. However, in social science research, no statistical analysis method is perfect. Multilevel statistical model for those nonlinear data which can’t convert to linear, such as the non-stationary process in the time series and some desultorily data has its limitations. The Chinese professor, DENG Julong, has constructed new data with accumulation method and explored the theory which has been successfully adopted to practice. This paper gives the modeling methods of GM(1,1) after introducing the scientific background of the gray theory, and then briefly introduces the classification of grey model and application, and finally, points out the own limitations of gray model.On the basis of the multilevel statistical model and the grey model, the paper presents a new model, called accumulating multilevel statistical model AMM(1,1) and explains the principle detailedly. On this basis, this paper makes innovations on the original model theory and establishes a new improved model. Finally the theory, with the practice use the improved AMM(1,1) model in the rainfall forecast of Yangtze river delta region. And comparing prediction effect of the AMM(1,1) model and the improved AMM(1,1) model can conclude that the improved model of the prediction accuracy is higher. There is a new method of the rainfall forecast. |
PDF Full Text Request