分析與預測波羅的海運價指數波動之趨勢---應用模糊時間序列法

Abstract

不定期國際散裝海運市場接近於完全競爭市場，運價波動劇烈，市場價格由市場供給與需求間所形成之均衡狀況而決定，使得散裝海運企業面臨著極大經營風險與不確定性；企業期望尋求避險之策略，波羅的海運價指數BDI(Baltic Dry Index)因應而生。根據不同船型以及航線在航運市場上的重要程度和所占權重構成的綜合性指數，可藉由BDI的變化，看出散裝航運市場景氣之起伏。

過往研究多使用傳統時間序列模式，如灰色理論模式以及ARIMA模式預測BDI走勢；本研究提出以模糊時間序列模式對時變性BDI進行預測，針對歷史資料的模糊性和模糊相關性的問題進行探討，將誤差項視為隸屬於相同的模糊語意變數但隸屬度不同所造成的結果，進而建構短期預測模式，希冀能提供不同以往的預測方式。

影響模糊時間序列的預測準確度有多種屬性，論域界定、區間分隔、模糊集合定義、資料模糊化、模糊關係建立、預測轉換、反模糊化函數之選定，都是預測模式中須注意之處；本研究結合以及修改文獻中較合理的屬性選定，建立單變量以及Type-2模式，用於預測BDI之走勢，進而分析本研究模式之準確度以及模式穩定性，發現屬性修改可得到良好的結果，值得後續研究。

International tramp bulk marine market is a perfectly competitive market. The market price depends on the equilibrium of supply and demand of the market. It makes the enterprise to confront enormous operation risk and uncertainty. Baltic Dry Index (BDI) provides a channel for hedge, which according to the important degrees and weight of different ship size and route. It is a comprehensive index. We could see the variation of bulk marine market condition by BDI.

Past research use ARIMA model or gray theory model to forecast the tendency of BDI. We propose a fuzzy time series model to forecast the time-variant BDI. We discuss the fuzziness and the fuzzy relation of history data. Finally, we construct two short-term models. We hope it could be a different forecast method from past research.

There are many attributions could affect accuracy in fuzzy time series model: define the universe of discourse, partition interval, define fuzzy sets, fuzzify data, establish fuzzy logical relationship and defuzzify forecast data. We combine the reasonable attributions from reference and build one-variable model and Type-2 model. We find the accuracy and robustness are improved. The method is worth further study.