A New Discordancy Test on a Regression for Cylindrical Data
Document Type
Article
Publication Date
1-1-2018
Abstract
A cylindrical data set consists of circular and linear variables. We focus on developing an outlier detection procedure for cylindrical regression model proposed by Johnson and Wehrly (1978) based on the k-nearest neighbour approach. The procedure is applied based on the residuals where the distance between two residuals is measured by the Euclidean distance. This procedure can be used to detect single or multiple outliers. Cut-off points of the test statistic are generated and its performance is then evaluated via simulation. For illustration, we apply the test on the wind data set obtained from the Malaysian Meteorological Department.
Keywords
Circular-linear, cylindrical data, k-nearest neighbour’s distance, outlier
Divisions
MathematicalSciences
Funders
Ministry of Higher Education, Malaysia under the Fundamental Research Grant Scheme No: FP037-2014B
Publication Title
Sains Malaysiana
Volume
47
Issue
6
Publisher
Penerbit Universiti Kebangsaan Malaysia