A second order model for cylindrical data
Author:
- 1 April 2000
- journal article
- research article
- Published by Taylor & Francis in Journal of Statistical Computation and Simulation
- Vol. 66 (1) , 51-65
Abstract
Modeling cylindrical data, comprised of a linear component and a directional component, can be done using Fourier series expansions if we consider the conditional distribution of the linear component given the angular component. This paper presents the second order model which is a natural extension of the Mardia and Sutton (1978) first order model. This model can be parameterized either in polar or Cartesian coordinates, and allows for parameter estimation using standard multiple linear regression. Characteristic of the new model, how to compare the adequacy of the fit for first and second order models, and an example involving wind direction and temperature are presented.Keywords
This publication has 4 references indexed in Scilit:
- An extension to modeling cylindrical variablesStatistics & Probability Letters, 1997
- Statistical Analysis of Circular Data1993
- Robust Locally Weighted Regression and Smoothing ScatterplotsJournal of the American Statistical Association, 1979
- A Model for Cylindrical Variables with ApplicationsJournal of the Royal Statistical Society Series B: Statistical Methodology, 1978