| 研究生: |
林宇軒 Lin, Yu-Hsuan |
|---|---|
| 論文名稱: |
結合超越機率準則之小樣本厚尾適應型無母數管制圖 A Heavy-Tail-Adaptive Distribution-Free Control Chart Incorporating Exceedance Probability Criteria for Small Samples |
| 指導教授: |
李俊毅
Lee, Chung-I |
| 學位類別: |
碩士 Master |
| 系所名稱: |
管理學院 - 統計學系 Department of Statistics |
| 論文出版年: | 2026 |
| 畢業學年度: | 114 |
| 語文別: | 中文 |
| 論文頁數: | 145 |
| 中文關鍵詞: | 超越機率準則 、厚尾分布 、尾指數 、無母數管制圖 、小樣本 |
| 外文關鍵詞: | exceedance probability criterion, heavy-tailed distribution, tail index, nonparametric control chart, small sample |
| 相關次數: | 點閱:4 下載:0 |
| 分享至: |
| 查詢本校圖書館目錄 查詢臺灣博碩士論文知識加值系統 勘誤回報 |
本文探討小樣本情境下結合超越機率準則(exceedance probability criterion, EPC)之無母數管制界限建構問題。原先方法以分數次序統計量與 Hutson(2002)對數型尾端外插為基礎,雖可於有限樣本下建立 EPC 管制界限,但當資料呈現厚尾特性時,常無法充分反映尾端行為,進而影響達成涵蓋率與管制界限之穩定性。因此,本文以 Parzen(1979)所提出之尾指數觀點為理論基礎,發展一套可依尾端厚度及訊息調整尾端外插方法並進一步納入一般化點繪位置參數,以提升小樣本下 EPC 管制界限之涵蓋率表現與實作彈性。
本文由密度分位函數之尾端模型建立左右尾尾指數之估計方式,並據以推導對應之廣義外插分位數公式。後續透過四種厚尾分布進行分析,比較本文方法與原有對數型尾端外插法於不同樣本數設定下之達成涵蓋率表現。模擬結果顯示,相較於 Hutson(2002)對數型外插,廣義外插法在四種厚尾分布下整體皆可提高達成涵蓋率,其中 Cauchy、Pareto(1) 與 Tukey(-0.5) 分布下改善幅度較為明顯,達成涵蓋率約可提升數個百分點至近二成;而在 t(4) 分布下,由於其尾端型態相對接近對數型外插所對應之情形,兩者表現較為接近。此外,為進一步處理小樣本下達成涵蓋率仍可能低於預設 EPC 下名目涵蓋率之問題,本文引入一般化點繪調整並引入事前設定之名目涵蓋率與樣本數建立自適應之選取規則,透過適當之點繪調整,可使達成涵蓋率達到所設定之名目涵蓋率。最後,本文亦將所提方法應用於半導體製程資料,建構符合 EPC 概念之無母數管制界限。模擬研究結果顯示,本文方法所得之上管制界限較能反映厚尾資料之尾端風險,並有助於降低既有方法可能產生之過度誤警。整體而言,本文提出之方法可作為小樣本厚尾情境下無母數管制圖建構之一種可行改良方向,除具理論意義外,亦可提升實務製程監控之穩定性與應用價值。
This study investigates the construction of nonparametric control limits under the exceedance probability criterion (EPC) for small samples. Existing methods based on fractional order statistics and Hutson's logarithmic tail extrapolation may perform poorly when the underlying distribution is heavy-tailed. To address this issue, we develop a heavy-tail-adaptive extrapolation method based on Parzen's tail index and further incorporate a generalized plotting-position parameter β to improve coverage performance and practical flexibility.
Simulation results show that the proposed method generally achieves better attained coverage than the existing logarithmic extrapolation approach under heavy-tailed distributions. In the semiconductor manufacturing application, the proposed method yields upper control limits that better reflect tail risk and may reduce excessive false alarms. Overall, the proposed procedure provides a practical improvement for EPC-based nonparametric control chart construction in small-sample heavy-tail settings.
Albers, W., Kallenberg, W. C. M., and Nurdiati, S. (2004). Parametric control charts. Journal of Statistical Planning and Inference, 124(1):159-184.
Andrews, D. (1973). A general method for the approximation of tail areas. The Annals of Statistics, pages 367-372.
Cheng, C. (1995). Uniform consistency of generalized kernel estimators of quantile density. The Annals of Statistics, 23(6):2285-2291.
Csörgő, M. and Révész, P. (1978). Strong approximations of the quantile process. The Annals of Statistics, 6(4):882-894.
Cunnane, C. (1978). Unbiased plotting positions - a review. Journal of hydrology, 37(3-4):205-222.
Fisher, R. A. and Tippett, L. H. C. (1928). Limiting forms of the frequency distribution of the largest or smallest member of a sample. Mathematical Proceedings of the Cambridge Philosophical Society, 24(2):180-190.
Gnedenko, B. V. (1943). Sur la distribution limite du terme maximum d'une serie aleatoire. Annals of Mathematics, 44(3):423-453.
Holan, S. H. and McElroy, T. S. (2010). Tail exponent estimation via broadband log density-quantile regression. Journal of Statistical Planning and Inference, 140(12):3693-3708.
Hutson, A. D. (2002). A semi-parametric quantile function estimator for use in bootstrap estimation procedures. Statistics and Computing, 12(4):331-338.
Jones, M. (2002). On fractional uniform order statistics. Statistics & probability letters, 58(1):93-96.
Meeker, W. Q., Hahn, G. J., and Escobar, L. A. (2017). Statistical intervals: A guide for practitioners and researchers. John Wiley & Sons.
Parzen, E. (1979). Nonparametric statistical data modeling. Journal of the American Statistical Association, 74(365):105-121.
Parzen, E. (2004). Quantile probability and statistical data modeling. Statistical Science, 19(4):652-662.
Staudte, R. G. (2017). The shapes of things to come: Probability density quantiles. Statistics, 51(4):782-800.
Stigler, S. M. (1977). Fractional order statistics, with applications. Journal of the American Statistical Association, 72(359):544-550.
Yang, H.-J. and Li, C.-I. (2025). Nonparametric control limits incorporating exceedance probability criterion for statistical process monitoring with commonly employed small to moderate sample sizes. Quality Engineering, 37(1):145-161.