Predictive Analysis of Taiwan Industry Product Index with Time Series

• Main Authors: Guo-shuo Zeng, 曾國碩
• Other Authors: Yu-Pin Lin
• Format: Others
• Language: zh-TW
• Published: 2009
• Online Access: http://ndltd.ncl.edu.tw/handle/83624405775151627640

碩士 === 真理大學 === 財經研究所 === 97 === In this thesis, we propose to construct two time series models to forecast the Taiwan industrial productive index. These time series models are Univariate Season Autoregressive Integrated Moving Average (SARIMA) based on industrial productive index and multivariate Linear Transfer Function (LTF) based on economical variables. The data is divided into two periods. The one period is January 1995 to June 2007 that we set consist of monthly observations for industrial productive index, the other is from July 2007 to June 2008 that we use SARIMA forecasting and LTF forecasting to calculate 12 forecasted values. We adopt MSE、MAE and MAPE to compare the prediction performance of these two methods. We show that multivariate LTF can is structured precisely better than univariate SARIMA. In comparative process, we fund there is the best goodness-of-fit value, but there must be not the best predictive value.