Keywords: Many regressions, cluster analysis, multidimensional scaling, mixture models
Many data sets, especially Big Data sets, are complex, in the sense that the observations are generated by many different mechanisms. This talk describes a procedure which iteratively extracts simpler component structures from complex data. The ideas are illustrated in the context of regression, cluster analysis, and multidimensional scaling. Portions of the work are based upon ideas implicit in earlier research by Ed Wegman.