There are also many other options such as building models for each group, easilty traverse those objects and produce easy to use ouput, etc…performing operations by group id(s) is powerful and easy to use.

This is a great package!

One difference is that plyr is built to take input from any one of 3 formats and output anyone of 4 ways. Another difference is that plyr does not have the functionality to do the equivalent of the ‘melt’ command from reshape.

Here’s my heuristic on when to use which package:

1) if normalizing / de-normalizing a data frame is the main purpose: reshape
2) Summarizing a data frame: either package
3) taking data from one type of object, doing a split, apply, combine operation and outputting in another object: plyr

I also have a number of summary type things that could be done in either. I tend to do those with plyr so I don’t have to remember both syntax sets. I generally only use reshape for #1 above.

I hope that helps.

The reshape package: http://had.co.nz/reshape/