Simulation of errors in linear regression: an approach based on fixed percentage area (Q1887228)
From MaRDI portal
 | This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use this page instead for the normal view: Simulation of errors in linear regression: an approach based on fixed percentage area |
scientific article; zbMATH DE number 2118518
| Language | Label | Description | Also known as |
|---|---|---|---|
| English | Simulation of errors in linear regression: an approach based on fixed percentage area |
scientific article; zbMATH DE number 2118518 |
Statements
Simulation of errors in linear regression: an approach based on fixed percentage area (English)
0 references
24 November 2004
0 references
A linear regression model with non-Gaussian errors is considered. The author considers two approaches to the comparison of Gaussian and non-Gaussian errors. In the first approach the errors with equal variances are compared (FV). In the second one the quantiles of a fixed level are equated for both models (FPA). It is claimed that FPA is more consistent for heavy-tailed distributions. Comparison of the power of tests for error normality (Shapiro-Wilk, Anderson-Darling, Bowman-Shenton) for FV and FPA is made via simulations.
0 references
non-Gaussian error
0 references
quantile
0 references
heavy tail
0 references
0.7479183077812195
0 references
0.7245758175849915
0 references
0.7217172980308533
0 references
0.721412718296051
0 references
