GAP(廣義相關圖) 是一套探索式的視窗統計分析軟體(exploratory data analysis, EDA)。主要目的是提供使用者以全矩陣式的視覺化方法,互動式地觀測資料的相關性結構及特性。我們開發了多種資料排序及群集分析方法,例如: 橢圓排序(R2E),階層式分群法,K均值法等等。以色彩及不同呈現條件,展現原始資料矩陣及兩個相關矩陣的結構,具有許多目前市面上統計分析軟體所沒有的功能。以廣義相關圖為發展主軸,我們亦針對數個重要的生物醫學資料形態進行結構分析,以模組方式,創造出適合各種資料形態之不同資訊視覺化環境,以利未來選擇或發展合適之統計方法與合理之生物與數學假設,為生物及醫學特徵資料提供一個多樣化之群集分析與資訊視覺化工作環境。目前此套軟體已開放下載試用,並在一些國際學術研討會及教學中演示,獲得許多迴響。

Title: GAP: A graphical environment for matrix visualization and cluster analysis
Abstract: GAP is a Java-designed exploratory data analysis (EDA) software for matrix visualization (MV) and clustering of high-dimensional data sets. It provides direct visual perception for exploring structure of a given data matrix and its corresponding proximity matrices, for variables and subjects. Various matrix permutation algorithms and clustering methods with validation indices are implemented for extracting embedded information. GAP has a friendly graphical user interface for easy handling of data and proximity matrices. It is more powerful and effective than conventional graphical methods when dimension reduction techniques fail or when data is of ordinal, binary, and nominal type.
Keyword: Clustering; Data visualization; Exploratory data analysis; Heat map; Java; Matrix visualization; Seriation; Software; Statistical graphics.[Ongoing Work]

 

Current version: v0.2.5 build 20080315.
History: 2004/04~2005/06: VisualEnvir; 2005/06~present: GAP
HomePage: http://gap.stat.sinica.edu.tw/Software/GAP
Mirror Site: http://www.hmwu.idv.tw/GAPSoftware  

 

2231180
Today
This Week
This Month
All days
515
3065
22179
2231180

Your IP: 18.209.66.87
2024-03-28 10:38