LOWLAD: A locally weighted \(L_ 1\) smoothing spline algorithm with cross validated choice of smoothing parameters (Q1893530)

The authors explore to extend the nonparametric estimator known as the \(L_1\) or least absolute deviations (LAD) smoothing spline to situations where a locally ``windowed'' robust smoother, similar to the lowest smoother of \textit{W. S. Cleveland, S. J. Devlin} and \textit{E. Grosse} [J. Econometrics 37, No. 1, 87-114 (1988; M.R. 89a:62157)] is called for. Complementing this windowed smoothing spline is a cross validation (CV) score which allows the data at hand to select the proper degree of smoothing within the local window, and to select the proper size window. An algorithm scheme, coupling the computation of the local LAD smoothing spline with the CV-dictated optimal choice of the smoothing parameters is presented. For background material on smoothing splines see for example \textit{G. Wahba} [Spline models for observational data, CBMS-NSF Regional Conference Series in Applied Mathematics, 59, Philadelphia, PA: SIAM, XII (1990; Zbl 0813.62001)].