Identification of the components of the drinking water model: The use of ARMA models

Abstract

In this paper, we consider the modeling of time series corresponding to the methodology of Box and Jenkins for building forecasting modelling from ARMA (Auto Regressive Moving Average)processes. This involves the analysis and forecasting of drinking water production data. The purpose of this study is to modelling a time series of drinking water production data according to its past and present values in order to determine the adequate ARMA process by the principle of parsimony. Using the Eviews for Windows software, the results show that the exploration of monthly drinking water consumption data provided by the Biskra Management Authority (A.D.E) during the period from January2009 to January2016, Revealed characteristics such as the non-stochastic non-stationarity of the drinking water production series.