The Identification of Outliers in Wrapped Normal Data by using Ga Statistics
Mohammad Illyas Sidik1, Adzhar Rambli2, Zamalia Mahmud3, Raiha Shazween Redzuan4, Nur Huda Nabihan Md Shahri5
1Mohammad Illyas Sidik, Faculty of Computer and Mathematical Sciences, Centre of Statistical and Decision Sciences Studies, University Technology MARA, Shah Alam, Selangor, Malaysia.
2Adzhar Rambli, Faculty of Computer and Mathematical Sciences, Centre of Statistical and Decision Sciences Studies, University Technology MARA, Shah Alam, Selangor, Malaysia.
3Zamalia Mahmud, Faculty of Computer and Mathematical Sciences, Centre of Statistical and Decision Sciences Studies, University Technology MARA, Shah Alam, Selangor, Malaysia.
4Raiha Shazween Redzuan, Faculty of Computer and Mathematical Sciences, Centre of Statistical and Decision Sciences Studies, University Technology MARA, Shah Alam, Selangor, Malaysia.
5Nur Huda Nabihan Md Shahri, Faculty of Computer and Mathematical Sciences, Centre of Statistical and Decision Sciences Studies, University Technology MARA, Shah Alam, Selangor, Malaysia.
Manuscript received on 01 February 2019 | Revised Manuscript received on 07 February 2019 | Manuscript Published on 13 February 2019 | PP: 181-188 | Volume-8 Issue- 4S February 2019 | Retrieval Number: DS2857028419/2019©BEIESP
Open Access | Editorial and Publishing Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open-access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: This paper focuses on identifying outliers in the wrapped normal distribution. It is commonly found and when it is dealing with circular data, the existing of outliers will increase several problems.We will be using the existing statistics, the 𝑮𝒂 statistics to identify a single and patch of outliers in the wrapped normal data. A Monte Carlo simulation will be carried out to generate the cut-off point’s value. The power performance of the discordancy test in circular data has been investigated. The increment of the contamination level, 𝝀, large value of concentration parameter, 𝝆 and large sample size, 𝒏 will increase the performance of the outlier detection procedures. In addition, the result shows that the statistics performs well in detecting a patch of outliers in the data. As an illustration a practical example is presented by using the wind direction in Kota Bharu station. As conclusion, the 𝑮𝒂 statistics successfully detect outlier presence in this data set.
Keywords: Circular Data, Outliers, 𝑮𝒂 Statistics, Wrapped Normal Distribution, Monte Carlo Simulation, Wind Direction.
Scope of the Article: Cryptography and Applied Mathematics