With grouping classifications, in contrast, one’s starting point is a collection of individual units of some given type—whether people, houses, shopping centers, etc. One then selects some given variable—one that can be measured on a continuous scale (e.g., ratio, interval, percent, etc.)—and then determines for each individual (or a sample of same) its numerical ―value on that variable. One can then either ―feed these numbers into a grouping program (univariate) or—more typically—do the same for a series of other variables, and then feed the resulting numbers into a grouping program (multivariate). To read more, download the full PDF here.