using the plyr package in R

Abstract. The merits of the plyr package are illustrated.

1. Introduction.

The base R system provides lapply() and related functions, and the package plyr provides alternatives that are worth considering. It will be assumed that readers are familiar with lapply() and are willing to spend a few moments reading the plyr documentation, to see why the illustration here will use the ldply() function.

The test task will be extraction of latitude (and then both latitude and longitude) from the section dataset in the oce package. (Users of that package may be aware that there is a built-in accessor for doing this, so results can easily be checked.)

2. Methods

First, load the data


Next, find latitudes using lapply

lat <- unlist(lapply(section[["station"]], function(x) x[["latitude"]]))

Next, find latitudes with ldply

lat <- ldply(section[["station"]], function(x) x[["latitude"]])

3. Results

The reader can check that the results match, although ldply() returns a data frame, not a vector as in the first method. Tests of speed indicate such a difference too small to be of much interest, at least in this case

microbenchmark(ldply(section[["station"]], function(x) x[["latitude"]])$V1, 
    unlist(lapply(section[["station"]], function(x) x[["latitude"]])))
## Unit: milliseconds
##                                                               expr   min
##        ldply(section[[&quot;station&quot;]], function(x) x[[&quot;latitude&quot;]])$V1 18.99
##  unlist(lapply(section[[&quot;station&quot;]], function(x) x[[&quot;latitude&quot;]])) 18.36
##     lq median    uq   max neval
##  20.26  20.56 21.02 36.05   100
##  19.71  19.93 20.64 63.18   100


4. Discussion

Since ldply() returns a data frame, it is more flexible than unlist(), which returns a vector. For example, the following creates a data frame with columns for lat and lon:

latlon <- ldply(section[["station"]], function(x) c(x[["latitude"]], x[["longitude"]]))

A station plot is produced as follows.

if (!interactive()) png("plyr.png", unit = "in", 
    res = 150, width = 7, height = 7, pointsize = 10)
mapPlot(coastlineWorld, projection = "orthographic",
     orientation = c(20, -40, 0))
mapPoints(latlon$V2, latlon$V1, pch = "+", 
    cex = 1/2, col = "red")
if (!interactive())
Map of station locations, produced by extracting locations using a function from the plyr package

Map of station locations, produced by extracting locations using a function from the plyr package

5. Conclusions

The effort of learning how to use the plyr package is likely to pay off in more flexible code, particularly because of the use of data frames in that package. On this theme, note that the author of plyr is developing a similar package called dplry, which centres more closely on data frames and offers many new features; see for a blog item introducing dplyr.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s