English  |  正體中文  |  简体中文  |  全文筆數/總筆數 : 83114/112007 (74%)
造訪人次 : 21745697      線上人數 : 885
RC Version 6.0 © Powered By DSPACE, MIT. Enhanced by NTU Library IR team.
搜尋範圍 查詢小技巧:
  • 您可在西文檢索詞彙前後加上"雙引號",以獲取較精準的檢索結果
  • 若欲以作者姓名搜尋,建議至進階搜尋限定作者欄位,可獲得較完整資料
  • 進階搜尋
    政大機構典藏 > 商學院 > 統計學系 > 學位論文 >  Item 140.119/119716
    請使用永久網址來引用或連結此文件: http://nccur.lib.nccu.edu.tw/handle/140.119/119716


    題名: 以一致性進行特徵選取
    Feature Selection Based On Kappa Statistics
    作者: 沈妤芳
    Shen, Yu-Fang
    貢獻者: 周珮婷
    林怡伶

    沈妤芳
    Shen, Yu-Fang
    關鍵詞: 機器學習
    特徵選取
    Kappa
    一致性
    Machine learning
    Feature selection
    Kappa
    Consistency
    日期: 2018
    上傳時間: 2018-08-29 15:47:32 (UTC+8)
    摘要: 本文針對特徵選取做主要研究對象,在機器學習中分類是很重要的一環,從過去到現在有許多方法和文獻採用特徵冗餘的方式消除彼此間相對不重要進而保留其他相關性差異大的變數,這裡將針對特徵配適後的模型做分類後,對於特徵之間的關係在這裡選用Kappa一致性為一指標,再透過分類結果相似的特徵組合起來作為特徵選取的方法,與其他如:Random Forest、ReliefF、mRMR和建構在Symmetric Uncertainty的特徵選取演算法下做比較,對於準確度和變數子集合數量發現都有不錯的效果。
    Feature selection plays an important role in supervised learning by eliminating irrelevant features and improving classification results. The current study proposed a feature selection method based on Kappa statistics to select consistent features. SVM was used as a single-variable classifier and Kappa statistics was computed from the fitted results as an indicator of relationship between features. The proposed method was compared with other methods such as, Random Forest, ReliefF, mRMR, and Symmetric Uncertainty based method. The results showed that the proposed method can effectively select important features and achieve stable prediction performance.
    參考文獻: Acid, S., De Campos, L.M., & Fernandez , M. (2011). Minimum redundancy maximum relevancy versus score-based methods for learning Markov boundaries. 2011 11th International Conference on Intelligent Systems Design and Applications, pp. 619-623.
    Breiman, L. (2001). Random Forest. Machine Learning, 45(1), pp. 5-32.
    Carletta, J. (1996). Assessing agreement on classification tasks: The kappa statistic. Computational Linguistics, 22(2), pp. 249-254.
    Cohen, J. (1960). A Coefficient of Agreement for Nominal Scales. Educational and Psychological Measurement, 20(1), pp. 37-46.
    Durgabai, R.P.L. (2014). Feature Selection using ReliefF Algorithm. International Journal of Advanced Research in Computer and Communication Engineering, 3(10).
    Fushing, H., & McAssey, M. (2010). Time, temperature, and data cloud geometry. PHYSICAL REVIEW E, 82,061110.
    Genuer, R., Poggi, J.M., & Tuleau-Malot, C. (2010). Variable selection using Random Forests. Pattern Recognition Letters, 31(14), pp. 2225-2236.
    Kononenko, I., & Robnik-Sikonja, M. (2003). Theoretical and Empirical Analysis of ReliefF and RReliefF. Machine Learning, 53, pp. 23-69.
    Kononenko, I., Robnik-Sikonia, M., & Pompe, U. (1996). ReliefF for estimation and discretization of attributes in classification, regression and IPL problems. Artificial Intelligence: Methodology, Systems, Applications: Proceedings of AIMSA'96, pp. 31-40.
    McHugh, M. (2012). Interrater reliability: The kappa statistic. Biochemia Medica, 22(3), pp. 276-282.
    Peng, H., Long, F., & Ding, C. (2005). Feature Selection Based on Mutual Information: Criteria of Max-Dependency, Max-Relevance, and Min-Redundancy. IEEE Transactions On Pattern Analysis And Machine Intelligence, 27(8), pp. 1226-1238.
    Sim, J., & Wright, C. (2005). The Kappa Statistic in Reliability Studies: Use, Interpretation, and Sample Size Requirements. Physical Therapy, 85, pp. 257-268.
    Singh, B., Kushwaha, N., & Vyas, O.P. (2014). A Feature Subset Selection Technique for High Dimensional Data Using Symmetric Uncertainty. Journal of Data Analysis and Information Processing, 2, pp. 95-105. doi:http://dx.doi.org/10.4236/jdaip.2014.24012
    描述: 碩士
    國立政治大學
    統計學系
    105354014
    資料來源: http://thesis.lib.nccu.edu.tw/record/#G0105354014
    資料類型: thesis
    DOI: 10.6814/THE.NCCU.STAT.016.2018.B03
    顯示於類別:[統計學系] 學位論文

    文件中的檔案:

    沒有與此文件相關的檔案.



    在政大典藏中所有的資料項目都受到原著作權保護.


    社群 sharing

    著作權政策宣告
    1.本網站之數位內容為國立政治大學所收錄之機構典藏,無償提供學術研究與公眾教育等公益性使用,惟仍請適度,合理使用本網站之內容,以尊重著作權人之權益。商業上之利用,則請先取得著作權人之授權。
    2.本網站之製作,已盡力防止侵害著作權人之權益,如仍發現本網站之數位內容有侵害著作權人權益情事者,請權利人通知本網站維護人員(nccur@nccu.edu.tw),維護人員將立即採取移除該數位著作等補救措施。
    DSpace Software Copyright © 2002-2004  MIT &  Hewlett-Packard  /   Enhanced by   NTU Library IR team Copyright ©   - 回饋