Forecasting Coconut Offered at Colombo Coconut Auction: ARIMA Modeling. Time Series Analysis

      Аннотация: Fresh coconuts are put up for the sale in Colombo coconut auction which is conducted by Coconut Development Authority. The present study was carried out with the objectives to identify the time series pattern of offered coconut quantity and selecting the best fitted model for short term and long term forecasting in Colombo coconut auction. The time series analysis methods i.e. ARIMA, Moving Average, Single and Double Exponential Smoothing were used to forecast the offered coconut quantity and the time series plots were used to identify the time series patterns like seasonal and non-seasonal,etc. in offered coconut quantity. ARIMA (0, 0, 1) (1, 1, 0) was fitted as the best ARIMA forecasting method for short term and long term forecasting. With using test data set, it was found that ARIMA (0,0,1) (1,1,0) has given the predicted values which are more close to the actual offered coconut quantities. The lowest Mean Absolute Percentage Error (MAPE) value (10.55%) was recoded in ARIMA (0, 0, 1) (1, 1, 0). It proves that ARIMA (0, 0, 1) (1, 1, 0) was the best fitted forecasting method among the other tested methods.

Ключевые слова: ARIMA, Offered coconut quantity, Time series methods