One Solution to the ‘stringsAsFactors’-Problem Or: Hell-Yeah there is HELLNO

December 23, 2015
By

(This article was first published on Automated Data Collection with R Blog - rstats, and kindly contributed to R-bloggers)

Base R's stringsAsFactors default setting is supposedly the most often complained about piece of code in the whole R infrastructure. A search through the source code of all CRAN packages in December 2015 (Link) resulted in 3,492 mentions of stringsAsFactors. Most of the time these explicit mentions where found within calls to data.frame() or as.data.frame() and simply set the value to FALSE.

The hellno package provides an explicit solution to the problem without changing R itself or having to mess around with options. One could use e.g.: options("stringsAsFactors" = FALSE) to re-set the global default behavior. Instead hellno tackles the problem from another direction, namely by providing alternative implementations of data.frame() and as.data.frame(). Those re-implementations are in fact simple wrappers around base R's very own data.frame() and as.data.frame() with stringAsFactors option set to HELLNO – which in turn equals to FALSE and gives the package its name.

Some info material and crediting for 'hellno' as catch phrase – thanks Clint?:

Using hellno interactively

Using the package is simple – load it, note the message indicating masking two base functions and code on – from now on no type conversion will take place within data.frame() and as.data.frame():

# options(repos = c(CRAN = "https://cran.rstudio.com"))
# install.packages("hellno")
library(hellno)
## 
## Attaching package: 'hellno'
## 
## Die folgenden Objekte sind maskiert von 'package:base':
## 
##     as.data.frame, data.frame
df2 <- data.frame(a=letters[1:3])
class(df2$a)
## [1] "character"

Using hellno for package development

While using hellno in interactive R is nice, in fact its real strength is that it can be imported when writing packages. Once imported stringsAsFactors=FALSE will be the default for all uses of data.frame() and as.data.frame() within all package functions BUT NOT OUTSIDE OF IT.

Thus it provides a way to ease programming while also ensuring that package users can still choose which flavor of stringsAsFactors they like best.

Let us see how this works following a little example. Again, let us start with loading hellno package:

library(hellno)
data.frame(a=letters[1:2])$a 
## [1] "a" "b"

As shown before, character vectors are not transformed to factor when hellno is loaded.

We unload hellno again to start clean.

unloadNamespace("hellno")

Now we install the hellnotest package from GitHub and load it. The hellnotest package imports hellno and therefore its function hellno_df() will not convert character vectors to factor while functions outside the packages scope will not be affected:

if( !("hellnotests" %in% installed.packages()) ){
  devtools::install_github("petermeissner/hellnotests")
}

library(hellnotests)

While all functions within the package use hellno's alternative implementations:

hellno_df
## function () 
## {
##     data.frame(a = letters[1:3])$a
## }
## 

… and hence for them character vector conversion does not happen anymore:

hellno_df()
## [1] "a" "b" "c"

… functions outside the package (like data.frame() from the base package) are not affected at all:

data.frame(a=letters[1:2])$a 
## [1] a b
## Levels: a b

Summing it up

  • Using hellno interactively makes the change of the default setting very explicit.
  • Writing packages with hellno does not change outside behavior.
  • R is Rsome.

Have fun and …

Discussion

If you have thoughts/ideas on the "stringsAsFactors"-problem, e.g. you do not like this solution because … I herewith open the issues section of the package's GitHub repository for general discussion of the theme and related stuff. I am very much interested on what you think.

To leave a comment for the author, please follow the link and comment on their blog: Automated Data Collection with R Blog - rstats.

R-bloggers.com offers daily e-mail updates about R news and tutorials on topics such as: Data science, Big Data, R jobs, visualization (ggplot2, Boxplots, maps, animation), programming (RStudio, Sweave, LaTeX, SQL, Eclipse, git, hadoop, Web Scraping) statistics (regression, PCA, time series, trading) and more...



If you got this far, why not subscribe for updates from the site? Choose your flavor: e-mail, twitter, RSS, or facebook...

Comments are closed.

Sponsors

Never miss an update!
Subscribe to R-bloggers to receive
e-mails with the latest R posts.
(You will not see this message again.)

Click here to close (This popup will not appear again)