透過您的圖書館登入
IP:3.21.76.0
  • 學位論文

以機器學習方法建立產品開發問題處理優先度評估模型

Building Issue Priority Assessment Models for Product Development based on Machine Learning Methods

指導教授 : 劉敦仁

摘要


在現今筆記型電腦研發過程中,處理研發問題的優先度評估一直是依靠有經驗的研發人員主觀判斷,同樣的問題組合,可能會因為負責的研發人員不同而得出不同的優先度判斷,主觀判斷難以量化也導致管理人員管理上的困難,本文希望透過深度學習方法來提出一個問題處理優先度評估模型,使用D公司產品開發問題管理系統匯出的資料做為研究資料集,特徵選取則利用資料集中的結構化資料,以及透過TFIDF方法找出文字資料的語詞特徵作為模型建立及訓練用的特徵值。 模型建立上使用SVM、隨機森林、類神經網路,三種分類模型進行分析比較,從中找出最佳的模型。 本文結果顯示在對問題處理優先度評估時,以使用類神經網路的分類方式表現最佳且召回率達85% 關鍵詞:優先度評估、機器學習

並列摘要


In today's laptop research and development process, the priority evaluation of handling research and development problems has always relied on the subjective judgment of experienced R&D personnel. The same problem combination may result in different priority judgments because of different responsible R&D personnel. Judgment is difficult to quantify and also leads to management difficulties for managers. This article hopes to propose an Issue Processing Priority Assessment Model through machine learning methods, using the issue data exported from the D Company’s Issue Management System as a research dataset. Feature selection uses structured data and text data as the feature values for model building and training. Random forests, neural networks and SVM three classification models are used for analysis and comparison to find the best model. The results of this study show that when evaluating the priority of issue processing, the classification method using neural networks performs best and the recall rate is 85%. Keyword: Priority assessment, Machine learning

參考文獻


摘要 i
Abstract ii
致謝 iii
目錄 iv
表目錄 vi

延伸閱讀