Would variables that are derived from other variables via summary statistics (e.g., mean, mode, standard deviation) be classified as interval or ratio type? Or neither?
2
-
2$\begingroup$ It depends on the statistics. But one wonders why it would even matter. $\endgroup$– whuber ♦Jan 1 at 15:51
-
$\begingroup$ @whuber in general, it most probably won't. however, some datasets in the well-known online ML repos contain such variables and i was wondering how to classify them (so more a question out of curiosity). my own thoughts (should have included them in the question): for mean and mode it depends on the variable that is used to derive these values (variable is of interval type -> mean and mode are of interval type, variable is of ratio type -> mean and mode are of ratio type). and for standard deviation it's always ratio (since there is a 'true' 0). Don't know if this is correct. $\endgroup$– keezarJan 1 at 17:30
Add a comment
|