[Hadley Wickham][1]. He has several exploratory data analysis projects on Github that you can look at (e.g., "data-baby-names"), and given the awesomeness of ggplot2/plyr/reshape, I have a default (but admittedly blind) trust in his best practices, particularly with respect to his own packages. Plus, you get an early heads up on other projects he's working on! [1]: https://github.com/hadley