Using ARIMA Model to forecast production of kharif sweet potato in Odisha

Abstract

Sweet potato, which is frequently referred to as a small farmers crop, is a major vegetable crop of Odisha. In India, Odisha is producing the most sweet potatoes. During the kharif season (June to August), sweet potatoes are grown as rain fed crop. A timely and accurate forecast of the area and production of such major vegetable crop is helpful for making agricultural policy decisions and giving nutrients to the population as the sweet potato occupies a key place among vegetable crops in Odisha. The objective of the present research is to predict the area, yield, and production of kharif sweet potato in Odisha by applying the most widely used forecasting model, ARIMA model. ACF and PACF plots and secondary data on the area, yield, and production of kharif sweet potato were collected from 1970–1971 to 2019–20 to fit the models that were determined to be appropriate. The best fit model was chosen based on the importance of the estimated coefficients, model diagnostic tests, and model fit statistics. By refitting the model with data from the most recent 4 years, 3 years, 2 years and 1 year, as well as by making one step ahead forecasts for the years 2016–17 to 2019–20, and the best fit model was cross-validated. The models with the best fits for the area, yield, and production of kharif sweet potatoes were found to be ARIMA (1,0,0) with constant model, ARIMA (0,1,1) without constant model, and ARIMA (0,1,1) without constant without constant. The forecast values show that the area, yield, and production of kharif sweet potatoes in Odisha will remain constant in future years, regardless of variation in the lower and upper class interval of the forecast values.