Nonlinear interval regression analysis based on spline fuzzy model with interval coefficients

Interval regression analysis is an effective method of handling uncertain and imprecise data. In this paper, a kind of spline fuzzy model is designed for modeling interval data. Spline function is chosen as the membership function, and the coefficients of this model are taken as interval-valued numbers. A target function based on approximation errors and specificity is proposed to improve the quality of estimated interval data. Accordingly, gradient descent algorithm is employed to tune these interval weights. Some numerical simulations are carried out to validate the effectiveness of the proposed spline interval regression model.