人類使用資源有限,當人口增長而資源需求量增加時,資源價格就會上漲,除了石油以外,金屬和農產品之漲價最明顯。銅一直是重要的金屬,家用器皿,電線,電池,許多電器產品的重要材料。隨著技術的進步,採用天然銅製造的生產工具,已被證明是不足以應付需求。相對於其他金屬,銅已經被用於為作為交易介質(貨幣)和商品在全球市場上有價值的商業產品。 台灣銅短缺的問題,我們需要從國外每年都進口了相當數量的銅。由於銅在扮演著台灣經濟增長的重要作用。本研究選取七個變化的因素包括:(1)生產和消費的銅(2)LME銅價(3)未來對銅的需求(4)LME倉庫銅庫存(5)原油價格波幅(6)主要國家匯率和(7)其他一些因素。自從2013年,由於國內銅產量增加,國內供應量一直高於需求量,台灣修訂銅出口政策,而且增加銅的出口。本研究目的是進行基於兩個預測模型多元回歸和灰色理論的預測。 建立預測銅價波動模式的灰色預測模型,和多元回歸預測模型,會用三種預測評價指標:RMSE,MAPE和泰爾的U,比較與實際值預測值的精確度,從而找出哪個模型的預測是對未來的投資更好。 本研究發現在MAPE以及Theil’s U這兩個預測績效評估指標的評估,這兩模式皆為優良的預測模式,灰預測模式在指標的評估仍是優於複迴歸模式。
Human resources were limited, when population growth and increased demand for resources, resource prices will rise, except for oil, metals and agricultural prices of the most obvious. Copper is an important metal, household utensils, wires, batteries, important materials for many electrical appliances. With advances in technology, use of natural copper manufactured means of production, has proven to be inadequate to meet demand. Compared with other metals, copper has been used as a medium of transaction (currency) and valuable commodity in the global market for commercial products. Taiwan copper shortage problem, we need to import from abroad every year a considerable number of copper. Copper played Taiwan an important role for economic growth. This study, seven varying factors include: (1) production and consumption of copper, (2) the LME copper price, (3) the future demand for copper, and (4) LME Warehouse copper stocks, (5) crude oil price volatility (6) exchange rates (7) number of other factors. Since 2013, increase in domestic copper production, domestic supply is higher than demand. Taiwan revised export policy, but also increase copper exports. This study is based on prediction model of two multiple regression and grey system theory forecast. Grey prediction model for establishment of prediction model of copper price fluctuations, and Mult-regression model, use the evaluation to three pointers: RMSE, MAPE and Theil’s U, comparison of predicted values with the actual values of accuracy. This study found that in MAPE and Theil’s U both predictive performance evaluation assessment of pointer, these two models are excellent model, assessment of grey prediction model in the pointer is still superior to regression models.