Volume no :15, Issue no: 2, June (2016)

Journal of Statistics: Advances in Theory and Applications

COMPARATIVE STUDY OF SEASONAL AUTOREGRESSIVE INTEGRATED MOVING AVERAGE AND FOURIER MODELS IN MODELLING RAINFALL DATA: A CASE OF AKWA IBOM STATE

Author's: Anthony Effiong Usoro
Pages: [85] - [102]
Received Date: March 8, 2016; Revised May 18, 2016
Submitted by:
DOI: http://dx.doi.org/10.18642/jsata_7100121653

Abstract

This study compares seasonal autoregressive integrated moving average model with Fourier series model in modelling seasonal data. Rainfall data in Akwa Ibom State are used for the study. The data are collected from January 2007 to December 2014, from Directorate of Statistics, Ministry of Economic Development, Akwa Ibom State. Investigation reveals that Fourier model gives good results as well as SARIMA, but smoothens the estimates better than the SARIMA model. This is evident in the Akaike Information Criterion (AIC). The findings may be different if another set of seasonal data is used, but it is admissible that Fourier model gives better results when modelling Akwa Ibom State rainfall data. Hence, this paper recommends Fourier series models for the analysis and forecasts of most series that are characterised by sinusoidal behaviour, particularly, Akwa Ibom State rainfall data.

Keywords

SARIMA model, Fourier model, rainfall data.