The wavelet identification for jump points of derivative in regression model

A method is proposed to detect the number, locations and heights of jump points of the derivative in the regressive model n(i) = f(xi (i)) + epsilon (i), by checking if the empirical indirect wavelet coefficients of data have significantly large absolute values across fine scale levels. The consistency of the estimators is established and practical implementation is discussed. Some simulation examples are given to test our method. (C) 2001 Elsevier Science B.V. All rights reserved.Statistics & ProbabilitySCI(E)5ARTICLE2167-1805