TOPSIS Method And Its Application To Time Series Forecast Model

Time series has been widely used in economic,political,financial,management,military and other fields.The prediction of time series,in which can explore the inherent laws and estimate the trend of the development of things,is a hot topic for scholars all over the world The evaluating indexes are used to determine the weight allocation of each single forecasting method in combination forecast.There are few papers on combination forecast and decision-making problems.Based on the existing achievements,this dissertation introduces TOPSIS method into combination forecast as follows(1)Considering the particularity of the stock index data,the data are preprocessed firstly,and the expansion of the common time series forecast method is proposed(2)Considering the question of the determination of the most optimal weight coefficient of combination forecast is MADM problem actually.The value and the n-order variance of the accuracy of the forecasts are looked upon as attributes.The induced value of IOWA operator can be calculated at time t based on approach degree.A combination forecast model is presented based on TOPSIS method and IOWA operator.A stock index example is illustrated to show that the method has effectiveness(3)The sensitivity analysis of the established model is pointed out to show that the method induced in this paper can embody more effective information.The generalization of the method is also discussed in the last of the dissertation.