FUZZY TIME SERIES FORECASTING WITH BELIEF MEASURE

Abstract

The profit of investment does not lie solely in the accuracy of prediction, but in the degree of belief as well. The greater the degree of belief is, the more capital the investors might venture, which results in more profit returns. On the contrary, under the condition of an accurate prediction, if the degree of belief is little, investors will not put in too much capital, which leads to limited profit. This study attempts to apply belief functions in explaining the prediction results of multivariate fuzzy time series, i.e. the degree of belief that the prediction model has for the prediction result. By utilizing multivariate fuzzy time series model, combining with two variables of closing price and volume of transaction in weighted stock price index, the author tries to predict the Taiwan weighted stock price index and estimate the degree of belief, which are trusted to be of great meaning for risk control and a better rate of return.