Families of discrete circular distributions with some novel applications
Kanti V. Mardia,Karthik Sriram
DOI: https://doi.org/10.48550/arXiv.2009.05437
2022-04-29
Abstract:Motivated by some cutting edge circular data such as from Smart Home technologies and roulette spins from online and casino, we construct some new rich classes of discrete distributions on the circle. We give four new general methods of construction, namely (i) maximum entropy, (ii) centered wrapping, (iii) marginalized and (iv) conditionalized methods. We motivate these methods on the line and then work on the circular case and provide some properties to gain insight into these constructions. We mainly focus on the last two methods (iii) and (iv) in the context of circular location families, as they are amenable to general methodology. We show that the marginalized and conditionalized discrete circular location families inherit important properties from their parent continuous families. In particular, for the von Mises and wrapped Cauchy as the parent distribution, we examine their properties including the maximum likelihood estimators, the hypothesis test for uniformity and give a test of serial independence. Using our discrete circular distributions, we demonstrate how to determine changepoint when the data arise in a sequence and how to fit mixtures of this distribution. Illustrative examples are given which triggered the work. For example, for roulette data, we test for uniformity (unbiasedness) , test for serial correlation, detect changepoint in streaming roulette-spins data, and fit mixtures. We analyse a smart home data using our mixtures. We examine the effect of ignoring discreteness of the underlying population, and discuss marginalized versus conditionalized approaches. We give various extensions of the families with skewness and kurtosis, to those supported on an irregular lattice, and discuss potential extension to general manifolds by showing a construction on the torus
Statistics Theory