REAP

Peter writes:

The result is a simple table with the columns being 'Data set name', 'Mean' and 'Variance', one row for each data set. These data could also be presented as a scatter plot of mean and variance, mean on one axis, variance on the other. Such a plot might show that although the means are similar between some of the data sets, the variances are not; the data have much more scatter in them but a low overall bias. One might run a simple clustering algorithm on these to determine which fit together best with which.