Hybrid Least-Squares Regression Modelling Using Confidence Bounds
One of the questions regarding bridging of soft computing and statistical methods is the (re-)use of information between the two approaches. In this context, we consider in this paper whether statistical confidence bounds can be used in the hybrid fuzzy least squares regression problem. By using the confidence limits as the spreads of the fuzzy numbers, uncertainty estimates for the fuzzy model can be provided. Experiments have been conducted in the paper, both on regression coefficients and the predicted responses of regression models. The findings show that the use of the confidence intervals as the widths of memberships gives successful results and opens new possibilities in system modeling and analysis.
KeywordsFuzzy Number Triangular Fuzzy Number Fuzzy Random Variable Fuzzy Regression Fuzzy Arithmetic
Unable to display preview. Download preview PDF.
- 1.Bardossy, G., Fodor, J.: Evaluation of Uncertainties and Risks in Geology. Springer, Berlin (2002)Google Scholar
- 3.Bowermann, B.L., O’Connell, R.: Linear Statistical Models: an applied approach. Duxbury Press, Boston (2000)Google Scholar
- 5.Chang, Y.H.O., Ayyub, B.M.: Hybrid least-squares regression analysis. In: Ayyub, B.M., Gupta, M.M. (eds.) Uncertainty Analysis in Engineering and Sciences. International Series in Intelligent Technologies, vol. 11, ch. 12. Kluwer Academic Publishers, Boston (1998)Google Scholar
- 13.Milton, J.S., Arnold, J.C.: Introduction to Probability and Statistics. McGraw-Hill, Singapore (1995)Google Scholar