Abstract:
This thesis is aimed to apply the nonlinear model to forecast the latex volume.
The model uses correlation of meteorological parameter data such as mean air
temperature, mean dew point temperature, mean vapour pressure, mean relative
humidity and etc. The data is input to the nonlinear model to forecast the latex volume.
Then, the model applies the Nonlinear Auto-Regressive with eXogenous input (NARX)
for forecasting the latex volume. The model uses the polynomial equation for predicting
the latex volume. The result of the analysis found that 6 initial maximum regressors and
40 initial maximum terms were the size of polynomial equation which caused the
minimum mean square error. This polynomial equation which is able to validate the
statistic principals has the independence of residuals, the normal distribution of
residuals and the dependence between the residuals and the input data with 95%
confidence intervals. When the researcher applied this polynomial equation to predict
the latex volume, the results are found that the errors are not over ± 5% compared with
the real volume of latex. In addition, this model can be developed to use to predict other
fields ; industrial works, service business, medical affairs and etc. Finally, this research
is useful for predicting the result accurately.