Analysis of Two Advanced Smoothing Algorithms.
Abstract
This thesis examines two smoothing algorithms which deviate from the classical method of using only one neighborhood size in the smoothing procedure. The Supersmooth algorithm uses three neighborhood sizes with local cross-validation in order to estimate an optimal neighborhood size. The Split Linear Fit algorithm uses any number of neighborhood sizes and computes a family of linear fits corresponding to each neighborhood size; the final smooth points are a weighted average of the linear fits. These two advanced smoothers are evaluated against the results produced by previously validated, commonly used smoothers and regression techniques. The measure of performance is the quality of the smooth curves and the value of the sum of squared residuals. Keywords: Test and evaluation; Computer programs; and Subroutines. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Sep 01, 1985
- Accession Number
- ADA162305
Entities
People
- Jose A. Vasquez Jr
Organizations
- Naval Postgraduate School