Full metadata record
DC FieldValueLanguage
dc.contributor.author陳潔馨en_US
dc.contributor.authorChen, Chieh-Hsinen_US
dc.contributor.author王昱舜en_US
dc.contributor.authorWang, Yu-Shuenen_US
dc.date.accessioned2015-11-26T00:57:13Z-
dc.date.available2015-11-26T00:57:13Z-
dc.date.issued2015en_US
dc.identifier.urihttp://140.113.39.130/cdrfb3/record/nctu/#GT070256621en_US
dc.identifier.urihttp://hdl.handle.net/11536/126997-
dc.description.abstract在現今的資訊時代中,將資料視覺化是一個必要的技術,視覺化可讓使用者快速並直覺的辨別大量資料的異同,找出特殊或異常的資訊再進而對這些資訊去做更深入的挖掘,許多研究者將資料視覺化成使用者易讀易懂的資訊,但大部分的研究分別著重在時間序列資料或是多變量資料的呈現,很少有結合兩者性質的研究方法,其原因在於會增加視覺化的複雜度,要如何將多變量的大量資訊呈現在有限的空間上是一個有難度的議題。為了克服這個問題,我們介紹了一個新的方法來對長時間的資料進行簡化,對於長時間龐大的資料,我們使用統計圖(histogram)來代表一個小範圍時間的資料,統計圖經過運算處理再進行視覺化後,得到統計圖的顏色與位置資訊,我們利用這兩個資訊來讓使用者辨別不同統計圖之間的相似度。統計圖對長時間資料的簡化大大節省視覺化的使用空間並允許所有統計圖同時顯示在螢幕上,使用者可以以全域的觀點在長時間資訊中快速直覺的找出特殊或有興趣的資訊,再進而對這些資訊作更詳細的挖掘與探索。多虧了互動式的設計,我們系統可以讓使用者很直覺的從全域觀點到局部觀點觀察與分析資料,在最後的案例研究中,我們使用了多組不同的長時間資料集的研究來證明我們方法的可行性。zh_TW
dc.description.abstractVisualization of time varying and multivariate data is essential. Trends or anomalies of data, or specific patterns that draw user attensions can be effectively identified. Previous methods mainly focus on the depiction of either time varying or multivariate data because the combination of these two properties would rapidly increase the visualization complexity. To prevent this problem, we introduce a new method to abstract longterm time varying multivariate data when depicting their properties. Specifically, a long dataset is divided into short segments and each of them is encoded to be a color coded bar. The bar shows the summarized information, say, histogram, of a short segment to briefly depict its property. Our system also aligns the bars based on their time coordinates and relative similarities to enhance readability. The presented visualization allows displaying a large amount of data while retaining as many details as possible. This space saving design can clearly reveal global evolutions and is helpful to data analysts. We show several case studies and the gained insights to demonstrate the feasibility of our system.en_US
dc.language.isozh_TWen_US
dc.subject統計圖zh_TW
dc.subject視覺化zh_TW
dc.subjectHistogramen_US
dc.subjectVisualizationen_US
dc.title多變量長時間視覺化系統zh_TW
dc.titleLongterm Multivariate Data Visualizationen_US
dc.typeThesisen_US
dc.contributor.department多媒體工程研究所zh_TW
Appears in Collections:Thesis