Modelling and Forecasting of Pigeonpea (Cajanus Cajan) production in Orissa using Arima Methodology

Abstract

An empirical study of modelling and forecasting time-series data of pigeonpea production in Orissa has been described in this paper. Box-Jenkins ARIMA time-series methodology was considered for modelling and forecasting Orissa’s pigeonpea production data (1971-72 to 2007-08). The augmented Dicky Fuller (ADF) test was applied to test stationarity in data set. Root Mean Square Error (RMSE), Akaike Information Criterion (AIC) and Bayesian Information Criterion (BIC) were used to identify the best model. Independence of errors were tested using “Run test” and “Durbin-Watson test”. The performance of fitted model was examined using Mean Absolute Error (MAE), Mean absolute percent prediction error (MAPE), Relative mean absolute prediction error (RMAPE), Root Mean Square Error (RMSE) and Theil’s Inequality Coefficients (TIC). ARIMA (1, 1, 0) model performed better among other models of ARIMA family for modelling as well as forecasting purpose. One and two-step ahead forecast value for year 2006-07 and 2007-08 for Orissa’s pigeonpea production was computed as 97.56 and 100.93 th. tonnes respectively.