Abstract
The problem of cubic spline smoothing of dependent data like time series and growth curves is addressed in this paper. Available statistical systems like S-PLUS (STATISTICAL SCIENCES, INC. 1991) and XploRe (XploRe SYSTEMS, 1992) do not provide appropriate algorithms. We propose a simple penalized least squares method with a number of computational advantages. It is called Dependent Error Regression Smoothing (abb. DERS) and implemented in a PC program under MS-Windows of the same name. The implementation comprises two techniques in an exploratory setting for smoothing parameter choice when the errors are serially correlated.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have