時間序資料的階層式聚類法

Hierarchical clustering for time-course data

蕭佑亘(Yu-Hsuan Hsiao)

指導教授 ： 楊敏生

中原大學/理學院/應用數學研究所/碩士(2013年)

https://doi.org/10.6840/CYCU.2013.00067

摘要

時間序資料是依事件或資料發生的先後次序排列的一群統計數據，時間序分析的目的在於觀察及分析過去的資料，時間序資料研究最基礎的問題在於從交易記錄中尋找有趣的順序性樣式(Sequential Pattern)，而時間序資料探勘是在時間序資料庫中尋找相似的循序樣式，或是於時間序資料庫中進行相似性的查詢；如在氣象資料中尋找符合某相似(循序)樣式的記錄、電信網路的警報分析、在疾病資料中探勘時間序樣式等。為此，我們想到使用聚類方法將相似的時間序資料進行聚類，藉此分析相似的循序樣式。在本文中我們使用相關係數加上階層法(hierarchical)來辨識時間序數據的圖形變化，並使用權重相關係數來修正相關係數使其分辨圖形的能力增加，並模擬了幾種常見的時間序類型數據進行分類，最後我們建議在聚類時間序數據上使用權重相關係數階層式聚類法。

關鍵字

階層聚類法 ； 聚類 ； 時間序資料

並列摘要

Time-course data is a group of statistic permuted according to the order of occurrence of events. The purpose of analyzing time-course data is to observe and analyze past data. The foundation of studying time-course data is to find interesting sequential patterns from transaction records. Meanwhile, time-course data mining is aimed at finding similar equential patterns in time-course database, or to conduct search according to similarity. For instance, finding similar (sequential) patterns in climate information, analyzing telecommunications network warnings, and conducting time-course mining in disease data. Due to the reasons above, we came up with applying clustering algorithm to group similar time- course data, and hence we can analyze similar sequential patterns. In this paper, we apply Hierarchical Clustering Algorithm combined with correlation coefficients to distinguish graph changing of time-course statistic. Also, we apply the Weighted Correlation Clustering Algorithm, which revises correlation coefficients to strengthen the graph-recognition ability. We have manipulated and then clustered some common time-course data, and we concluded that it's better to apply Weighted Correlation Clustering Algorithm on clustering time-course data.

並列關鍵字

Time course data ； hierarchical ； clustering

參考文獻

[1] M.S. Yang and K.L Wu, “A similarity-based robust clustering method,” IEEE Trans. Pattern Analysis and Machine Intelligence, vol. 26, pp. 434-448, 2004.

[2] K.L. Wu and M.S. Yang, “Alternative c-means clustering algorithms,” Pattern Recognition, vol. 35, pp. 2267-2278, 2002.

[3] J.C. Dunn, “A fuzzy relative of the ISODATA process and its use in detecting compact, well-separated clusters,” Journal of Cybernetics, vol. 3, pp. 32-57, 1973.

[4] E. Ruspini, “A new approach to clustering,” Information and Control, vol. 15, pp. 22-32, 1969.

[5] T. Kohonen, Self-Organizing Maps, Springer-Verlag, Berlin, 1995.