Abstract
Local regression, or loess, has become a popular method for smoothing scatterplots and for nonparametric regression in general. The final result is a "smoothed" version of the data. To obtain the value of the smooth estimate associated with a given covariate a polynomial, usually a line, is fitted locally using weighted least squares. This article presents a version of local regression that fits more general parametric functions. In certain cases, the fitted parameters may be interpreted in some way and we call them meaningful parameters. Examples are included that show how this procedure is useful for signal processing, physiological, and financial data.
Original language | English (US) |
---|---|
Pages (from-to) | 72-79 |
Number of pages | 8 |
Journal | American Statistician |
Volume | 55 |
Issue number | 1 |
DOIs | |
State | Published - Feb 2001 |
Keywords
- Circadian pattern
- Harmonic model
- Local regression
- Meaningful parameters
- Sound analysis
ASJC Scopus subject areas
- Statistics and Probability
- General Mathematics
- Statistics, Probability and Uncertainty