Using support vector machines to forecast the production values of the machinery industry in Taiwan

Abstract

The machinery industry is one of the most important exporting industries in Taiwan. The values of Taiwan’s machinery industry have been increasing continuously over the past years. Therefore, forecasting of production values is an essential issue for the machinery industry in Taiwan. Support vector machines (SVMs), a novel forecasting technique, have been successfully applied in solving non-linear regression and time series problems. In this paper, SVMs are employed to examine the feasibility in forecasting seasonal time series data of production values of Taiwan’s machinery industry. In addition, two other forecasting approaches, namely the seasonal time series autoregressive integrated moving average (SARIMA) model and general regression neural networks (GRNN), are used to compare the performance of forecasting. Experimental results show that support vector machines outperform SARIMA and the general regression neural networks in terms of forecasting accuracy.