碩士 / 國立臺灣海洋大學 / 航運管理學系 / 93 / Abstract
Taiwan is an island country, also a trade-based economic system. The import and export performance at the harbors contribute significantly to the harbor management. The development of a harbor requires considerable amount of capital, and the harbor has dedicated use after the development and unable to be used for other purposes. Therefore, insufficient understanding or inaccurate forecast of the future cargo quantity may result in waste of resources.
The loading and unloading of containers are the main income sourses of the harbor, thus, the forecast of container quantity is one of the crucial information for harbor planning, development, and management.
This study begins with literature review to examine the commonly adopted method for harbor forecast and the variables affecting the container quantity. The containers were classified into import containers, export containers, and transfer containers; export containers were used to forecast the future quantity in Keelung Harbor. The forecast methods on export containers include the classic decomposition method, trigonometric function regression method, and regression method, with seasonal dummy variables.
The data were collected from 1998 (January to December) to 2001 (January to December). The data from January 1998 to December 2000 were in-sample data; the data from January 2001 to December 2001 were out-of-sample data. After comparison by using Mean Absolute Error (MAE), Mean Absolute Percentage Error (MAPE) and Root Mean Square Residual (RMSR), We found that the classic decomposition method showed the best forecast capability, the regression method with the seasonal dummy variable showed better forecast capability than that of the trigonometric function regression method.
Based on the analysis, the empirical models generated by the forecast methods mentioned in this research were limited, thus, using the seasonal Autoregressive- Integrated-Moving average models, neutral network and combined forecast, or other new theories could effectively improve the forecast accuracy. Different forecast methods have their advantages and disadvantages. Finding suitable forecast model for different industries could be highly beneficial.
| Identifer | oai:union.ndltd.org:TW/093NTOU5301042 |
| Date | January 2005 |
| Creators | Hui-Ching Tsao, 曹慧菁 |
| Contributors | Chu Ching Wu, 朱經武 |
| Source Sets | National Digital Library of Theses and Dissertations in Taiwan |
| Language | zh-TW |
| Detected Language | English |
| Type | 學位論文 ; thesis |
| Format | 54 |
Page generated in 0.0109 seconds