>... > I never figured out how I was supposed to reverse the > correlations in order to have a 'distance' metric that would > work. Did you have some concrete advice from somewhere? > > Since the results I could get on correlations did not look > nearly as meaningful to me as what Factor Analysis > gave me, I decided to forget about MDS. >
MDS is not intended as a modeling method that helps you to establish analytical relationships. It is more a visualization tool for preliminary data analysis. It can help you, for instance, to answer questions like: Are there any clear clusters in a dataset? Does a data point have many similar data points in a dataset? Are there many isolated outliers in a dataset? You can probably answer these questions with statististics, but with MDS it is more direct and visual. The value and strength of MDS is its mapping feature which visualizes abstract high dimensional data. > Has somebody cared enough about MDS to update the > computer programs? It's long been my impression that > 'marketing' was using MDS. From google, it also seems > like MDS sometimes is included in the tools of data mining. That's right, MDS is more appropriated for areas where conventional analytical modeling have failed. For self-promotion purpose please see our software VisuMap at http://www.visumap.net which combines traditional MDS, PCA, clustering method with modern navigation user interface. VisuMap basically turns any numerical table ( or any distance matrix) into map. cheers, James X. Li . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
