軟體設計以GAP架構為核心,分析方法以維度縮減(Dimension Reduction)和群集分析(Cluster Analysis)方法為主。發展互動式的2D 及3D 的Scatterplot 來觀察降維後的資料結構。已實作的方法有奇異值分解(SVD)、主成份分析(PCA)、切片逆迴歸法(SIR)、多維度尺度法(MDS) 及ISOMAP。各個圖形的資料點可以相互連結並選取。目的是提供使用者更方便的EDA工具。 

 

 

jDRCluster is a Java-based tool for the dimension reduction techniques and cluster analysis. The methods include singular value decomposition (SVD), principle components analysis (PCA), sliced inverse regression (SIR), sliced average variance estimation (SAVE), principal Hessian direction (phd), multidimensional scaling (MDS) and ISOMAP. The output can be saved and plotted on the 2D or 3D scatterplots. The most important feature of the jDRCluster in the future will be the linking among graphics produced by different dimension reduction methods. [Ongoing Work]

 

Current version: v0.1 build 20070520.
History: 2004/06~2007/05: JDRViewer; 2007/05~present: jDRCluster

 

 

2231653
Today
This Week
This Month
All days
988
3538
22652
2231653

Your IP: 44.203.235.24
2024-03-28 22:21