Abstract
This paper presents robust estimators for binary and multinomial circular logistic regression, where a circular predictor is related to the response. An extensive Monte Carlo Simulation Study clearly shows the robustness of proposed methods. Finally, three numerical examples of Botany, Crime and Meteorology illustrate the application of these methods to Life and Social Sciences. Although in the Botany data the proposed method showed little improvement, in the Crime and Meteorological data an increment up to 5\% and 4\% of accuracy, respectively, is achieved.
Publisher
Universidad Nacional de Colombia
Reference25 articles.
1. Abuzaid, A. H. & Allahham, N. R. (2015), 'Simple circular regression model assuming wrapped cauchy error', Pakistan Journal of Statistics 31(4).
2. Al-Daffaie, K. & Khan, S. (2017), Logistic regression for circular data, in 'AIP Conference Proceedings', vol. 1842, AIP Publishing LLC, p. 030022.
3. Alshqaq, S. S., Ahmadini, A. A. & Abuzaid, A. H. (2021), 'Some new robust estimators for circular logistic regression model with applications on meteorological and ecological data', Mathematical Problems in Engineering 2021.
4. Ashby, M. P. (2019), 'Studying crime and place with the crime open database: Social and behavioural scienes', Research Data Journal for the Humanities and Social Sciences 4(1), 65-80.
5. Bell, K. (2008), Analysing Cycles in Biology and Medicine-a practical introduction to circular variables & periodic regression, Razorbill Press, St. John's.