Modelling Circular Time Series with Applications

Abstract

Circular time series modelling has posed a big challenge to many researchers due to the nature of observations. This article traces different circular time series models and the methods adopted in the development. It goes further to illustrate the use of these methods and the resultant circular time series models of wind direction data. The application involved hourly data on wind direction for the two major seasons (wet and dry seasons) in Nigeria. Training and development of the models are done using R codes and SPSS version 24. The results shows that score-driven models based on von Misses distribution is the best model using some error statistics and goodness of fit tests.