Modeling Data under Non-additive Measure-a Theoretical Frame and Some Applications

摘要

In this paper, we study how to estimate the nonlinear multi-regression model based upon fuzzy integral (either Choquet integral or (S)ipo(s) integral), and then propose a method of transforming this non-linear model into the linear one in real number field. Further, we simplify this method of nonlinear transformation to facilitate application. When we employ our model to analyze financial data, we find that both the goodness of fit and prediction ability is greatly improved.