For a school course I and a partner developed a GUI in R designed to enable exploration of data via visualization of hierarchical clustering and correlation of cluster partitions with external metadata. The key features were the ability to load in a distance matrix (most GUI-based clustering programs require feature vector input), and the ability to dynamically subset the data via the GUI built using a user-provided meta data file. I didn't think this was sufficient to publish a package, but it did seem like it could make for a good foundation. I so far have not come across other software that provides all of these features. I was hoping to get initial feedback in terms of whether anyone thought this, with or without certain enhancements, might be worthwhile. For a sense of what the existing tool looks like and what it does: http://mason.gmu.edu/~tgillett/R_Cluster_GUI/RClusterGUI.html
We intend to extend the tool to enable feature vector input, additional forms of visualization beyond cluster dendrogram, multidimensional scaling of distance matrix input, internal and external cluster statistics (e.g. Davies-Bouldin index, Rand index), random subsampling of data to account for the instability of clusters, and non-hierarchical clustering methods. Obviously a well-organized GUI is important, and we would seek out feedback and perhaps partner developers. I'd appreciate any direct feedback here, or suggestions of more appropriate forums for getting feedback and having a discussion on the topic. Thank you, Todd Gillette ______________________________________________ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
