Date of Award


Document Type

Thesis - ECU Access Only


Edith Cowan University

Degree Name

Doctor of Information Technology


School of Computer Science and Security


Faculty of Computing, Health and Sciences

First Supervisor

Jitian Xiao


Decision support systems (DSS) are used to support efficient and effective decision making in many circumstances such as forecasting, managing problems and analysing data. Although used globally in the agricultural, public, government and business sectors, DSS are yet to be adopted within the public agricultural rubber industry in Thailand (PARIT). As PARIT is a significant economic growth area in Thailand, it provides an applied setting for this study; a setting in which the objectives of the National Economic and Social Development Plan (Issue 10) for Thailand must apply. PARIT’s policy makers rely upon basic traditional statistical methods at present. If implemented, more modern forecasting techniques will offer better forecasts and facilitate innovation within PARIT.

After an extensive literature survey, this study utilised time series forecasting and DSS with artificial intelligence (AI) to produce an exemplar model for possible adaptation and application within PARIT. Experimentation was conducted to produce a proposed rubber latex price and national rubber production forecasting model which was based upon readily-available rubber data and software. Two techniques were utilised during the experimentation, namely non-neural and neural network (NN) training techniques. The experimentally-derived forecasts were compared with actual rubber latex price and national rubber production data to derive the best-fitting forecasting model.

The study confirmed that the non-NN training technique provided more accurate rubber latex price forecasts, as this technique was better able to deal with data sets with high levels of seasonal fluctuations. By contrast, the NN training technique was found to deliver more accurate national rubber production forecasts because there was little data fluctuation to cause errors or ‘noise’. This indicates that forecasts based on the NN training technique will be useful for implementation by policy makers in PARIT.

This proposed methodology can be recommended as a practical decision support tool for PARIT. However, the applications of particular forecasting techniques, and the findings from this study, are specific to PARIT. Further research over a longer period of time will be needed to judge more clearly how this forecasting model might be implemented in a wider range of agricultural industry sectors in Thailand and elsewhere.

LCSH Subject Headings

Decision support systems.

Artificial intelligence -- Agricultural applications -- Thailand.

Rubber industry and trade -- Thailand.


Paper Location