FORECASTING NATURAL RUBBER PRODUCTS IN THAILAND USING VARIOUS TIME SERIES TECHNIQUES

Abstract

The study attempted to determine suitable forecasting models to predict primary rubber products along the rubber supply chain in Thailand. Historical data used in this study were during the years 1999–2021. Two techniques were used to establish models, including stepwise multiple regression analysis and PCA. It was found that both techniques were suitable to predict quantities of natural rubber, smoked rubber sheets, block rubber, and concentrated latex. The MAPEs for these four predictions of both techniques ranged around 5%-8%. However, both techniques had not yet been able to sufficiently predict the quantities of mix and stock rubbers due to highly fluctuated raw data. Keywords : Rubber, Forecast, Principal Component Analysis, Time Series DOI： https://doi.org/10.35741/issn.0258-2724.58.3.24