Fuzzy regression without predefined functional form, or nonparametric fuzzy regression, is investigated. The two most basic nonparametric regression techniques in statistics, namely, k-nearest neighbor smoothing and kernel smoothing, are fuzzified and analyzed. Algorithms are proposed to obtain the best smoothing parameters based on the minimization of cross-validation criteria.
關聯:
Computers and Mathematics with Applications 38(3-4), pp.239-251
