Hi all, Does anyone know where to find more information on the "by" class? OR Does anyone know how to coerce an object of the "by" class into a data.frame containing the results of FUN and the values of the grouping variables? OR Does anyone know how to do the following: I have a big table with the results of many replications of an experiment and I want to summarize the medians according to the experimental parameters used. Columns 1-6 index the parameters used in that particular replication. Column 7 is just an index to the replication number. Column 8 is the variable I want to summarize I did: x=by(model.errors[, 8], model.errors[, 1:6], median) I now have an object of class "by" that seems to contain everything I need. If I just output "x" to the screen, I see all I need: the median for each unique combination of values in model.errors[, 1:6]. However, I only know how to access the information on the medians themselves by getting x[i] where i is the index of the median I want. The problem is: - How do I retrieve the combination of values in model.errors[, 1:6] for a given index "i". OR - How do I determine the correct index "i" for a particular combination of the values in model.errors[, 1:6]? Just to help the thinking: I bypassed this problem for the mean by using lm and predict. I could in principle do the same using the L1 loss instead of the L2 loss in lm by I do not know how to do that... This seems a rather convoluted solution for a problem "by" is meant to solve easily... Any help is appreciated. Thanks a lot, Guilherme Rocha