Loading...
|
Please use this identifier to cite or link to this item:
https://nccur.lib.nccu.edu.tw/handle/140.119/149026
|
| Title: | 快閃記憶體價格預測
Flash Memory Price Forecasting |
| Authors: | 高玉川
Kao, Yu-Chuan |
| Contributors: | 郭炳伸
Kuo, Biing-Shen
高玉川
Kao, Yu-Chuan |
| Keywords: | 快閃記憶體
多元迴歸
自迴歸
平均絕對誤差
平均絕對百分比誤差
NAND Flash Memory
Multiple Regression Model
Autoregressive Model
Mean Absolute Error (MAE)
Mean Absolute Percentage Error (MAPE) |
| Date: | 2023 |
| Issue Date: | 2024-01-02 15:20:08 (UTC+8) |
| Abstract: | 快閃記憶體價格佔產品總製造成本相當重要比例，能有效降低庫存就能降低企業營運風險。本研究藉由公開網站查詢或下載資料進行水準值分析，利用多元迴歸分析方式尋找快閃記憶體價格預測模型作為實驗組，同時以自迴歸模型AR(1)作為對照組，實際價格與預測價格平均絕對誤差( MAE ) 與平均絕對百分比誤差 ( MAPE ) 作為判斷預測價格精確度的指標。經分析後，3個月短期以多元迴歸模型精確度較佳，7個月中期以自迴歸模型AR(1)精確度較佳。
因變數本期值減上期值之差額轉換為本期因變數偏離值，其餘14個自變數也依此作法轉換為本期自變數偏離值，最後將所有偏離值視為待分析資料再次進行多元迴歸與自迴歸模型AR(1)偏離值分析後，短期與中期自迴歸模型AR(1)的精確度皆較佳，但差異並不大。若以水準值與偏離值分析做比較，結果顯現偏離值分析精確度較佳。
預測價格精確度固然重要，但是藉由價格漲跌趨勢來協助企業執行調節存貨風險決策，更有利於企業決策人員洞燭機先、領先整體市場的走勢，使得公司利潤最大化、風險極小化。本期預測價格與上期實際價格的相關性，可以用來輔助買進或賣出庫存的重要依據。預估本期價格與上期實際價格的漲跌，若與實際市場價格前後期的反應相同，預估模型會給予公司一個正確買入或賣出庫存處置訊息；反之，模型則給予一個錯誤處置建議。本研究藉由賦予上述4個預測模型適當權重，計算出短期加權預測價格，能有效輔助公司做出正確的決策。
The price of flash memory accounts for a very important proportion of the total product manufacturing cost. Effectively reducing inventory can reduce corporate operational risks. This study conducts level value analysis by querying or downloading data from public websites, and uses multiple regression analysis to find a prediction model for flash memory prices as the experimental group. At the same time, the autoregressive model AR(1) is used as the control group. The mean absolute error (MAE) and the mean absolute percentage error (MAPE) between the actual price and the forecast price are used as indicators to judge the accuracy of the forecast price. After analysis, the accuracy of the multiple regression model is better in the short term of 3 months, and the accuracy of the autoregressive model AR(1) is better in the mid-term of 7 months.
The difference between the current period's value of the dependent variable minus the previous period's value is converted into the current period's dependent variable deviation value. The remaining 14 independent variables are also converted into the current period's independent variable deviation value in the same way. Finally, all deviation values are regarded as data to be analyzed for multiple regression and autoregressive model AR(1) again. After analyzing the deviation values of regression and autoregressive models AR(1), the accuracy of the short-term and medium-term autoregressive models AR(1) is both better, but the difference is not large. Compare with level value and deviation value analysis, the results show that the accuracy of deviation value analysis is better.
Predicting price accuracy is important, but using price rise and fall trends to assist companies in making decisions to adjust inventory risks is more conducive to corporate decision-makers to be proactive and ahead of the overall market trend, maximizing company profits and minimizing risks. The correlation between the forecast price of this period and the actual price of the previous period can be used as an important basis to assist in deciding to buy or sell inventory. Estimating the rise and fall of the current period's price and the actual price of the previous period. If it reacts the same as the actual market price before and after, the prediction model will give the company a correct buy or sell inventory disposal suggestion; otherwise, the model will give an incorrect disposition suggestion. This study calculates the short-term weighted forecast price by giving appropriate weights to the above four forecast models, which can effectively assist the company to make correct decisions. |
| Reference: | 一、中文文獻
1.方文德 (2023)，半導體景氣循環現象之解析：以五吋廠為例，碩士，國立宜蘭大學。
2.王文箴 (2000)，台灣半導體製造業之景氣循環與經營績效分析，碩士，國立交通大學。
3.王瓊敏 (2000)，電腦關鍵零組件之價格預測模式，碩士，國立中央大學。
4.吳惠婷 (2012)，主流記憶體之二十年價格模式研究與驗證，碩士，國立中央大學。
5.吳福立 (2000)，DRAM價格變動模式之探討，碩士，國立交通大學。
6.巫宗穎 (2019)，快閃記憶體價格與營收預測，碩士，國立交通大學。
7.李小嫻 (2022)，記憶猶欣:考量需求動能之DRAM現貨價格預測，碩士，國立成功大學。
8.呂明權 (2012)，總體經濟因素對國際礦石價格之影響與預測，碩士，國立台北大學。
9.林亞嫺 (2014)，應用資料探勘於預測半導體價格之研究 - 以 DRAM 為例，國立中央大學。
10.林英志 (2010)，動態隨機存取記憶體產品價格預測研究，碩士，輔仁大學。
11.胡芳瑜 (2020)，總體經濟指標與美國十年公債殖利率之關聯性研究，碩士，國立政治大學。
12.柏洛賓 (2010)，利用多元回歸模型預測鐵礦石價格，碩士，國立台灣大學。
13.張家富 (2001)，DRAM價格模式研究，碩士，國立臺北大學。
14.郭寶謙 (2006)，DRAM月平均價格變動分析，碩士，國立中央大學。
15.陳新弘 (2019)，動態隨機存取記憶體需求、價格與營收預測，碩士，國立交通大學。
16.陳澤維 (2017)，臺灣蔬果批發市場價格預測模型，碩士，國立政治大學。
17.陳力行 (2007)，NAND型Flash價格與交運量預測在風險分析下之決策模式，碩士，國立中央大學。
18.黃憲廷 (2015)，美國之非製造業採購經理人指數與總體經濟變數之關係，碩士，國立中正大學。
19.黃子恒 (2013)，製造業採購經理人指數與總體經濟變數之關聯性研究－以美國為例，碩士，國立中正大學。
20.曾一峻 (2014)，我國動態隨機存取記憶體產品策略之研究-以W公司為例，碩士，國立臺灣科技大學。
21.葉麗貴 (2001)，DRAM季價格預測，碩士，國立交通大學。
22.蔡元哲 (2003)，動態DRAM價格之模擬分析---從DRAM產業及PC產業之研究，碩士，國立臺灣大學。
23.鄧啟民 (2009)，應用類神經網路預測快閃記憶體價格之研究，碩士，國立交通大學。
24.顏月珠 (2003)，商用統計學
二、英文文獻
1.D. Acemoglu and A. Scott (1997), Asymmetric business cycles: Theory and time-series evidence, Journal of Monetary Economics Vol. 40 Issue 3 Pages 501-533
2.J. Berman and J. Pfleeger (1997), Which industries are sensitive to business cycles ? , Monthly Labor Review Vol. 120 Issue 2 Pages 19-25
3.L. Jelinek (2018), Global semiconductor market trends, IHS Markit, May 2018
4.M. Lepselter and S. Sze (1985), DRAM pricing trends—the π rule, IEEE Circuits and Devices Magazine Vol. 1 Issue 1 Pages 53-54
5.R. Micheloni, L. Crippa, A. Marelli and G. Wong (2010), Market and applications for NAND Flash memories, Inside NAND Flash Memories 2010 Pages 1-18
6.Y. Tarui and T. Tarui (1991), New DRAM pricing trends: The Bi rule, IEEE Circuits and Devices Magazine Vol. 7 Issue 2 Pages 44-45 |
| Description: | 碩士
國立政治大學
經營管理碩士學程(EMBA)
111932107 |
| Source URI: | http://thesis.lib.nccu.edu.tw/record/#G0111932107 |
| Data Type: | thesis |
| Appears in Collections: | [經營管理碩士學程EMBA] 學位論文
|
Files in This Item:
| File |
Description |
Size | Format | |
|---|
| 210701.pdf | | 4073Kb | Adobe PDF | 1 | View/Open |
|
All items in 政大典藏 are protected by copyright, with all rights reserved.
|
