Statistical modeling and forecasting of Jammu and Kashmir apple prices

The Autoregressive integrated moving average model, developed by Box and Jenkins showed impressive and robust outcomes for forecasting market prices of agricultural products, finance and stock indices. In the present study, daily price of two apple varieties (super delicious and super american of Jammu and Kashmir, India) were studied for forecasting the market price. To address seasonality in the data series, a methodology called TBATS (Trigonometric, Box-Cox transform, ARMA errors, Trend, and Seasonal components) model was used. The TBATS model is helpful in capturing seasonality present at multiple periods. The different possible combinations of ARIMA model were determined and on the basis of minimum Akaike information criteria and Bayesian information criteria values and following the principle of parsimony, ARIMA (2, 1, 1) model for super american and ARIMA (3, 1, 6) model for super delicious were finally selected to forecast one month ahead. The out of sample data (test data) was used to compare the accuracy of the fitted models. The Mean Absolute Percentage Error (MAPE) was calculated as an accuracy measure of the fitted models and was found to be 0.05-0.08. Concluding that the fitted ARIMA model was 92-95% accurate for forecasting purpose in these data series.