Abstract

STATISTICAL MODELING OF THICK DUST HAYS HAZARDS IN SOKOTO METROPOLIS USING ARIMA MODEL

S. U. Ibrahim, M. O. Oladejo and U. L. Okafor

---

Abstract ARIMA model have been used in this research to model hazards values of a climatic factor such as thick dust hays in Sokoto metropolis under Box and Jenkins (1976) methodology. The climatic factor data series were obtained from Nigeria meteorological unit (NIMET) Sokoto as an annual average secondary data series for period of fifteen years (2002-2016) which were assumed to be stationary both in mean and variance over time. Hypotheses were tested in this research against time plots to test for certain statistical facts findings, i.e. test for auto correlations of errors and test for stationary process. The results of the tested hypotheses favored the climatic factor data series assumptions that the series were stationary over time. In the model parameters estimation of the climatic factor satisfied both stationarity and invertibility conditions i.e. at least one of the autoregressive parameter estimate should be negative and at least one of the moving average parameter estimate should be positive respectively, and the estimated ARIMA model parameters are: AR (1); 1.752, AR (2); -0.992, MA (1); 1.708, MA (2);-0.689 and MA (3); -0.053. The ARIMA model estimated parameters yielded ARIMA model order as ARIMA (2, 0, 3). The order of the best fitted ARIMA model was found to be adequate at all lags and auto correlations of errors were not significant. Keywords: Auto regressive, moving average and White noise process.

PDF