Fitting a Time Series Model

This study concentrates on fitting a time series §model for paddy production in Sri Lanka. There are §two main goals of time series analysis:(a) §identifying the nature of the phenomenon represented §by the sequence of observations,and(b)forecasting §First,I plot the series and examine the main §features of the graph, checking in particular if §there is a trend, a seasonal component, any apparent §sharp changes in behavior, or any outlying §observations. Then remove the trend and seasonal §component to get stationary residuals. To achieve §this goal it may some times be necessary to apply a §preliminary transformation to the data. A software §package ITSM96 for windows is used to analysis the §data, which was developed by P.J.Brockwell and §R.A.Davis.To fit a time series model to paddy §production data, first the non-stationary series was §converted to a stationary series. The ACF and PACF §graphs were used to find the order of the models. §Based on minimum FPE,AICC,BIC and Maximum Likelihood §statistic values, the several models such as MA,AR §and ARMA models were fitted. Finally these models §are used to forecast the paddy production up to year §2005.