Grey Regression Combination Forecasting Model Of S Type Fluctuation Sequence And Its Application

Prediction is the results of a study in the future to speculate in advance a kind of technology,and fully tap of inner development and evolution process.The paper based on the sequence forecast as the main research theme,through the establishment of?-Logistic model,improved Verhulst model,cubic exponential smoothing model and regression model to forecast the S type sequence,using example verification,compare and contrast,residual analysis and numerical simulation methods to solve the sequence prediction problem.First the model S series forecasting model,expounds the present situation of the research analysis.Then,introduces ?-Logistic model parameter estimation,through applied examples and numerical simulation show that using the ?-Logistic model forecast sequence,overcome the Logistic model's phenomenon of lag or advance prediction.Secondly introduces the grey system theory and grey prediction model,and using the GM(1,1)model respectively,the grey Verhulst model and generalized Verhulst model to forecast the S series problem,through the numerical simulation and residual analysis,the relative error is analyzed,the result of the forecast.Results show that the improved Verhulst model prediction prediction precision is better than another models.The third chapter introduces the cubic exponential smoothing method and parameter estimation;This paper cubic exponential smoothing method is used to predict the S type sequence,combined with Matlab are given cubic exponential smoothing method in S sequence specific application,the model's calculation is speed,the prediction error is small.The Logistic model's and the cubic exponential smoothing method's two aspects of theoretical deduction and theirs' practical application incomparison with,points out the advantages and disadvantages of each model,provide ideas for combination model.Finally introduces regression combination model,based on S type sequence grey combination of regression model,and comparing with other three kinds of single models to predict the results,through specific examples concluded that combined model prediction effect is better.