To Forecast Automobile Sale in Taiwan Using Adaptive Network-Based Fuzzy Inference System

• Main Authors: Chih-wei Tzeng, 曾治瑋
• Other Authors: Fu-Kwun Wang
• Format: Others
• Language: zh-TW
• Published: 2009
• Online Access: http://ndltd.ncl.edu.tw/handle/43579642446208903153

碩士 === 國立臺灣科技大學 === 工業管理系 === 97 === The demand forecasting of automobile sales is one of critical issues for national economic growth. Our model considers several variables such as current automobile sales quantity, coincident indicator, leading indicator, wholesale price index and income. Here, we only focus on new automobile sales in Taiwan. The data set is based on monthly sales which the data can be divided into three types of automobile sales. Thus, there are two levels in this study. First, we use the stepwise method to select most influential variables as our input variables. Then, we use three forecasting models: ARIMA (Autoregressive Integrated Moving Average Model), ANN（Artificial Neural Network）and ANFIS(Adaptive Network-Based Fuzzy Inference System) to forecast the automobile sales in each level. We use three forecasting models to compare their forecasting performance. The conclusion shows that the ANFIS model outperforms other two forecasting models. Key words: Automobile Industry, ARIMA, ANN, ANFIS, Time Series, Forecasting