數值型項目之循序樣式的變化研究 : 以降雨資料為例

機構典藏 > College of Business and Management > Graduate Institute & Department of Information Management > Thesis > Item 987654321/114517

Please use this identifier to cite or link to this item: https://tkuir.lib.tku.edu.tw/dspace/handle/987654321/114517

Title:	數值型項目之循序樣式的變化研究 : 以降雨資料為例
Other Titles:	A study of change detection for sequential patterns of numerical items : using rainfall data as an example
Authors:	張詮舜;Chang, Chuan-Shun
Contributors:	淡江大學資訊管理學系碩士班周清江
Keywords:	循序樣式;樣式改變;資料探勘;頻繁樣式;sequential pattern;Change detection;data mining;Frequent Pattern
Date:	2017
Issue Date:	2018-08-03 14:55:07 (UTC+8)
Abstract:	循序樣式探勘是從序列資料庫中，找出頻繁出現且有順序的樣式，通常這些樣式會再被轉換成先前所不知道的、有用的與有價值的資訊。過去相關研究大部分著重於演算法的效能以及為了蒐集到樣式的可用性而設定相關限制，如時間、效益等。以時間這個相關限制來看，過去研究較少考慮到在不同時間背景下，樣式會改變，使過去蒐集到的樣式之參考性下降。因此，本研究參考過去研究在循序樣式變化相關研究架構，提出一個能找出不同時段之循序樣式變化的方法，針對以強度分級的項目，把項目的強度差距納入考量，並改進在計算樣式差異度時因不符合三角不等式的原理造成的差異度不一致性。針對樣式變化分類，我們也提出一個新的分類方式，避免了過去研究重複分類的情形，最後，我們將此方法應用於中央氣象局的公開降雨資料，找出不同年間的降雨樣式的變化。 Sequential pattern mining is to extract frequent sequentially ordered patterns, which are usually transformed into previously unknown, meaningful, and useful information. Most of past frequent pattern mining researches focused on the effectiveness of algorithms. To increase the utilization of these patterns, they set constraints on time, benefits, and so on. In the aspect of timing, most past studies did not consider change of patterns in different periods. That caused the decline of utilization reference in the collected patterns. Therefore, based on previous researches on change detection of sequential patterns, for items representing different degrees of intensity, we propose a framework to extract change of their sequential patterns in different periods. We consider the intensity difference of the items, and remedy the inconsistency problem, caused by failing to conform the triangular inequality, of previous researches. As for pattern change classification, we propose a new classification method to avoid duplicate classification phenomena in previous researches. Finally, we apply this framework to the public rainfall data provided by Central Meteorological Bureau, to identify changes in the rainfall patterns over the years.
Appears in Collections:	[Graduate Institute & Department of Information Management] Thesis

Files in This Item:

File	Description	Size	Format
index.html		0Kb	HTML	145	View/Open

Loading...