以灰色預測模式預測企業用專線設備之研究–以“L”公司為例–

Abstract

台灣通訊產值由1996年八百多億到2007年的七千多億，而且，政府也大力的支持，推行各種國家級計畫如M-Taiwan，希望扶植國內的通訊設備商在國際上能如半導體般舉足輕重。

由於半導體技術的突飛猛進，促使通訊產業產生極大的變化，通訊產業已經由以前技術導向轉變成為需求導向，電信運營商不再受電信設備廠的限制轉而要求電信設備廠配合他們的需求生產設備，況且又遭逢2000年的電信泡沫及2008年的次級房貸，電信運營商大都減少資本支出，除非能夠有確定的商機，否則都處在觀望的態度。而半導體技術的進步，也使得進入障礙大大的降低，因而，使得電信市場處在百家爭鳴的狀況，使得，由以往的高利潤轉為低毛利。

因此，若是想要維持高毛利，除了研發新產品、開拓新市場，如何降低生產成本及符合客戶的交期也是相當重要。本研究藉由該公司產品2005年第一季至2008年第二季的銷售資料建立預測模型，並且以2008年第三季及第四季做為預測值與實驗值的比較分析。

研究結果發現，以四期灰色預測模型最佳、五期迴歸分析法次之四期指數平滑法最差，所以可以用四期灰色預測模型來做為企業用專線設備之後預測模型。

關鍵字：預測、灰色預測、通訊產業

From 1996 to 2007, the communication output product in Taiwan has grown from more than 80 billion to more than 700 billion. The Taiwanese government has devoted its full support to the communication industry and carried out plans at the national level, such as the M-Taiwan, in the hope that the communication companies in Taiwan may play an influential role internationally, as the semiconductor industry does. The rapid advancement of the semiconductor technology has brought a major change in the communication industry. The communication industry has shifted from a technology-driven to a needs-driven mode. That is to say, telecommunication companies are no longer restricted by the manufacturers and they now ask the manufacturers to develop products according to their needs. Moreover, most telecommunication companies have been trying to reduce their prime cost after the hit of telecome bubble in 2000 and subprime mortgage in 2008. Most telecommunication companies hesitate to enter the markets unless there are absolute business opportunities. However, the advancement of the semiconductor technology has greatly reduced the obstacles of entering the market. The result of this phenomenon is fierce competition and a great reduction on the profits. In order to maintain a high profit margin, it is important to develop new products, expand marketing territories, lower production cost, and also meet customers’ request on delivery dates. The present study attempts to establish a forecast model based on marketing data of the first quarter of 2005 to the second quarter of 2008 of the L Company. Moreover, the forecast value and experimental value, which will be based on the third and fourth quarter of 2008 will be analyzed and compared. The results of the study suggest that the grey prediction model with four periods ranked first, regression analyses with five periods ranked second, and the exponential smoothing forecast method with four periods came last. Therefore, the grey prediction model with four periods may be used for prediction in enterprises.

Keywords: Forecasting, gray prediction, and communication industries