dplyr Tutorial: verbs + split-apply

At a recent Saint Louis R users meeting I had the pleasure of giving a basic introduction to the awesome dplyr R package. For me, data analysis ubiquitously involves splitting the data based on grouping variable and then applying some function to the subsets or what is termed split-apply (typically split-lapply-apply). Having personally recently incorporated dplyr into my data wrangling workflows; I’ve found this package’s syntax and performance a joy to work with. My feeling about dplyr are as follows.

Data wrangling without dplyr.

Data wrangling with dplyr.

This tutorial features an introduction to common dplyr verbs and an overview of implementing split-apply in dplyr.

Some of my conclusions were; not only does dplyr make writing data wrangling code clearer and far faster, the packages calculation speed is also very high (non-sophisticated comparison to base).



The plot above shows the calculation time for 10 replications in seconds (y-axis) for calculating the median of varying number of groups (x-axis), rows (y-facet) and columns (x-facet) with (green line) and without (red line) dplyr.