Sains Malaysiana 52(7)(2023): 2139-2148

http://doi.org/10.17576/jsm-2023-5207-19

A Comparison between Two Discordancy Tests to Identify Outlier in Wrapped Normal (WN) Samples

(Perbandingan antara Dua Ujian Percanggahan untuk Mengenal Pasti Data Terpencil dalam Sampel Normal Balutan (WN))

NURISHA MOHD ZULKEFLI¹, ADZHAR RAMBLI ^1,* , MOHAMAD ISMETH KHAN AZHAR SUHAIMI¹, IBRAHIM MOHAMED² & RAIHA SHAZWEEN REDZUAN³

¹School of Mathematical Sciences, College of Computing, Informatics and Mathematics,

Universiti Teknologi MARA, 40450 Shah Alam, Selangor, Malaysia

²Institute of Mathematical Sciences, Universiti Malaya, 50603 Kuala Lumpur, Malaysia

³Centre for Foundation Studies in Science, Universiti Malaya, 50603 Kuala Lumpur,

Malaysia

Diserahkan: 17 Mac 2023/Diterima: 6 Julai 2023

Abstract

This study focuses on comparing the performance of the Robust Circular Distance (RCDU*) (simplified version) and A statistics in detecting a single outlier in the Wrapped Normal (WN) samples. Firstly, this study proposes a simplified version of RCDU statistic. Then, the paper generates the cut-off points for both statistics taken from WN samples via a simulation study. This study also evaluates the performance of both statistics using the proportion of a correct outlier detection. As a result, for a small sample size, the performance of RCDU* and A statistics do not have a huge difference. However, for a large sample size of n=250, A statistic performs slightly better than RCDU* statistic. As an illustration of a practical example, both statistics successfully detected one outlier present in the wind direction data at Kota Bharu station.

Keywords: Circular data; discordancy tests; outliers; wrapped normal distribution

Abstrak

Kajian ini memfokuskan kepada perbandingan prestasi Jarak Berkeliling Teguh (RCDU*) (versi ringkas) dan statistik A dalam mengesan satu data terpencil dalam sampel Normal Balutan (WN). Pertama, kajian ini mencadangkan versi ringkas statistik RCDU. Kemudian, kertas itu menjana titik potong untuk kedua-dua statistik yang diambil daripada sampel WN melalui kajian simulasi. Kajian ini juga menilai prestasi kedua-dua statistik menggunakan perkadaran pengesanan data terpencil yang betul. Akibatnya, untuk saiz sampel yang kecil, prestasi RCDU* dan statistik A tidak mempunyai perbezaan yang besar. Walau bagaimanapun, untuk saiz sampel yang besar n=250, statistik A menunjukkan prestasi yang lebih baik sedikit daripada statistik RCDU*. Sebagai ilustrasi contoh praktikal, kedua-dua statistik berjaya mengesan satu data terpencil hadir dalam data arah angin di stesen Kota Bharu.

Kata kunci: Data pekeliling; data terpencil; taburan normal balutan; ujian percanggahan

RUJUKAN

Abuzaid, A.H., Mohamed, I.B. & Hussin, A.G. 2009. A new test of discordancy in circular data. Communications in Statistics: Simulation and Computation 38(4): 793-809. https://doi.org/10.1080/03610910802627048

Collett, D. 1980. Data in circular outliers. Journal of the Royal Statistical Society 29(1): 50-57.

Fisher, N.I. 1993. Statistical Analysis of Circular Data. Cambridge: Cambridge University Press. https://doi.org/10.1017/CBO9780511564345

Hussin, A.G., Abuzaid, A.H., Ibrahim, A.N.I. & Rambli, A. 2013. Detection of outliers in the complex linear regression model. Sains Malaysiana 42(6): 869-874.

Jammalamadaka, S.R. & SenGupta, A. 2001. Topics in Circular Statistics. World Scientific.

Mahmood, E.A., Rana, S., Midi, H. & Hussin, A.G. 2017. Detection of outliers in univariate circular data using robust circular distance. Journal of Modern Applied Statistical Methods 16(2): 418-438.

Mohamed, I.B., Rambli, A., Khaliddin, N. & Hussin, A.G. 2016. A new discordancy test in circular data using spacings theory. Communication in Statistics - Simulation and Computation 45(5): 2904-2916. https://doi.org/10.1080/03610918.2014.932799

Rambli, A., Ibrahim, S., Abdullah, M.I., Hussin, A.G. & Mohamed, I. 2012. On discordance test for the wrapped normal data. Sains Malaysiana 41(6): 769-778.

*Pengarang untuk surat-menyurat; email: adzhar@uitm.edu.my