Tuesday, April 13, 2010

Excel v. R for Statistics

A previous post on the IEOR Tools blog led me to an article by Michael Milton on O'Reilly Answers about the relative merits of Excel and R for statistics. I tend to incline toward the position that Excel should never be used for statistics (beyond sample means, anyway). Milton links to a 2007 post by Frank Harrell on this theme; Harrell in turn links to other sources. Milton suggests that, beyond quick (and simple) computations, Excel would be the platform of choice for presenting statistical analysis. I'm not convinced; I would be more inclined to use a presentation package, embedding R-generated graphics, for that. Nonetheless, I'm happy to see R getting a bit more publicity, as well as a thoughtful discussion of the relative merits of R versus Excel. One option Milton missed is RExcel, an Excel add-in that provides access within Excel to R functionality. One of these days, time permitting, I need to play with it.