完整後設資料紀錄
DC 欄位語言
dc.contributor.author李恕毅zh_TW
dc.contributor.author劉敦仁zh_TW
dc.contributor.authorLi, Shu-Yien_US
dc.contributor.authorLiu, Duen-Renen_US
dc.date.accessioned2018-01-24T07:41:11Z-
dc.date.available2018-01-24T07:41:11Z-
dc.date.issued2017en_US
dc.identifier.urihttp://etd.lib.nctu.edu.tw/cdrfb3/record/nctu/#GT070463411en_US
dc.identifier.urihttp://hdl.handle.net/11536/141608-
dc.description.abstract半導體廠的整體產出是競爭的關鍵因素之一,本論文研究半導體廠過去一年機台各個參數(機台妥善率、機率效率、平均貨量、批貨專人處理時間、機台當機次數…等),對機台最後產出(move)的影響。參考半導體廠的九個參數,使用資料探勘方法中的線性迴歸、隨機森林及SVM進行時間序列的數值預測,藉以求得影響機台產出的最大關鍵參數,以期藉各項參數預測機台產出的資訊,提供產業決策方向。 經本研究的整理及比較,影響機台產出(move)最關鍵前三大參數為1.平均載貨量、2.機台效率、3.機台妥善率。同時比較資料探勘方法,本研究發現相對於線性迴歸與SVM,隨機森林的方法具有較佳的預測表現。zh_TW
dc.description.abstractThe semiconductor fab’s manufactured quantity is one of the key competition factors. This thesis studies the correlation with machine move and various parameters of the machine (eg. tool uptime, tool efficiency, average WIP, hold time, tool interruption count, etc.) of the semiconductor fab in the past year. Based on the nine parameters of the semiconductor fab, the linear regression, random forest and SVM data mining methods are used to predict the manufactured quantity of silicon wafer, so as to obtain the most important parameters which will affect the output of the tool move. The analysis can be used to provide industry decision-making direction. Through the evaluation and comparison, the most critical top three parameters that will affect the tool move are Average running WIP, Tool efficiency, and Tool uptime. The experimental result also shows that compared with the linear regression and SVM methods, the random forest method has a better predictive performance.en_US
dc.language.isozh_TWen_US
dc.subject資料探勘zh_TW
dc.subject數值預測zh_TW
dc.subject隨機森林zh_TW
dc.subject支援向量機zh_TW
dc.subjectData Miningen_US
dc.subjectPredictionen_US
dc.subjectRandom Foresten_US
dc.subjectSupport Vector Machineen_US
dc.title運用資料探勘方法預測矽晶圓製造產量zh_TW
dc.titleUsing Data Mining to Predict the Manufactured Quantity of Silicon Waferen_US
dc.typeThesisen_US
dc.contributor.department管理學院資訊管理學程zh_TW
顯示於類別:畢業論文