Abstract:
Nowadays, Durian is the most important exported fruit of Thailand. The export value of durian is approximately 144.87 million USD per year. The problem in this durian product has been over demanded in the season, which is June-August of every year, since the product has been brought out into the market simultaneously, causing durian growers sell their product lower than cost price. Therefore, in order to avoid the problem of durian exceeds the needs of consumers, the objective of this research is to design the forecasting model of the demand of durian in export markets. This research is investigated the demand of four kinds of durian: fresh durian, frozen durian, durian paste and durian chips in the next year. Kinds of model to forecast the demand have been employed in this research. The first model is applying Output models the four time series by Moving Average, Deseasonalised, Exponential Smoothing and Double Exponential Smoothing. The second model is applying Input models: which are Regression model and Artificial Neural Networks (ANNs) model. The forecast data from research models have been compared and analyzed by the Mean Absolute Percentage Error: MAPE. The lower MAPE value is the most accurate forecast model. The results of Output models reveal that the most accurate forecast model is Deseasonalised model. While, input models reveal that the most accurate forecasted model is Artificial Neural Networks (ANNs) model, the results from the forecasting models of four durian types in the domestic market and export market are assessed by using Linear Programming (LP) to active sales planning, in order to get the appropriate profit. The net profit has increased 16% accordingly.