應用類神經網路預測快閃記憶體價格之研究

Forecasting Nand Flash Prices by Neural Network

鄧啟民(Chi-Min Teng)

指導教授 ： 姜齊

國立交通大學/管理學院/管理學院碩士在職專班管理科學組/碩士(2009年)

https://doi.org/10.6842/NCTU.2009.00484

摘要

快閃記憶體Nand Flash 產業蓬勃發展了二十年，各式各樣的應用帶給人們更多的便利，並且改變了人類生活模式。1979年Sony 發明了第一台隨身聽，當初的人們正在為這神奇的新產品感到著迷的時候，誰能料想到三十年後，隨身聽不只能聽音樂，還能顯示歌詞，無限錄音，播放MTV，甚至有專屬音樂下載網站。照相機，攝影機也不再使用底片當成儲存媒介，不再為了儲存回憶而付出昂貴的成本。筆記型電腦可以更輕巧更超薄， 有著更快速更強大的功能，真正讓筆記型電腦更貼近人們的生活。這一切的演變都少不了這些身處Nand Flash產業中，各個努力不懈積極開發的廠商，假設地球上從沒出現過快閃記憶體，我想很多人都不能否認，這世界將會有另一個改變。

對於各大Nand Flash原廠與下游模組商來說，雖然Nand Flash 潛藏商機無限，但是也同時存在著高風險，因為佔銷售之產品總成本非常高比例的Nand Flash 晶片是一個隨時處於供給與需求不均衡狀態的產業，任何的稍不注意，便可能造成無法彌補的損失，這代表著在這Nand Flash 產業裡，有著幾項特性：”成本非常競爭”、”價格波動劇烈”、”明顯的景氣循環”、”市場型態隨著技術發展而隨時調整”。

透過類神經預測模型，規劃出一個可以預測未來Nand Flash 價格的方式，提供廠商在以經驗法則判斷的同時，能有多一些參考資訊。

關鍵字

匯率 ； 類神經網路 ； 價格預測 ； 股市指標 ； 快閃記憶體

並列摘要

The industry of Nand Flash has been growing for twenty years. Varieties of applications make human being more convenience and make changes in its life. Since 1979, Sony invented the first walkman in the world, while people crazy about such a amazing product, who can expect that after three decades: walkman is not only just a music player, it display lyrics when the song is playing, recording without limitation , playing MTV, even more it has own download web site. Camera and Video recorder use external memory device as the recording media, high cost for store memory is just a history. Laptop is simmer, more powerful features and easier to carry, makes it more close to people’s life. All the evolution brought to us by the people and enterprises who are dedicate in the field, I think everyone will admits that it is going to be a big different in the world if Nand Flash is never be shown.
Originals and module houses in the field of Nand Flash are all knows about the huge potential business opportunities, Even though, a high risky situation exists oppositely, because the Nand Flash is usual the components with high percentage share in cost is always in the unstable of supply and demand market situation, any inattentive may causes terrible loose. It means, there are some features in the field of Nand Flash：”extreme competitive in cost”、”pricing sharp fluctuations”、”obvious business cycle”、”market patterns are keeping changing by the technological development”.
This study try to work out a method through BPN for predicting the price of Nand Flash has not yet happened, to provide extra information except for rule of thumb to enterprises related to the industrial.

並列關鍵字

Exchange Rate ； Artificial Neural Network ； BPN ； Pricing predicting ； stock market indicator ； Nand Flash

參考文獻

(15) 張錫華，從DRAM產業發展經驗探討Flash Memory產業經營策略，元智大學，碩士論文，民國90年

(16) 陳力行，NAND型Flash價格與交運量預測在風險分析下之決策模式，國立中央大學，碩士論文，民國95年

(18) 陳永裕，動態隨機存取記憶體產業生命週期之研究，國立中央大學，碩士論文，民國94

(19) 陳立白，台灣快閃記憶卡製造廠商之競爭策略研究，國立清華大學，碩士論文，民國94年

(3) Donaldson, R.G. & Kamstra, M., Forecast Combining with Neural Networks, Journal of Forecasting, PP.49-61, 1996.