While we did our best to ensure that R for Everyone was error free, a few mistakes did slip into the book, both in code and text.  The following is a list of known mistakes that occur in some printings of the book.

Section 4.2 on page 37

When showing that the assignment operator <- can point to the right, the arrow was accidentally printed pointing to the left as:

> 3 <- z
> z
[1] 3

when it should be point to the right as:

> 3 -> z
> z
[1] 3

Section 6.3 on Page 75

The following line mistakenly said that setting stringsAsFactors to TRUE would ave processing time whereas it should be FALSE.

Original:

Again, setting this to TRUE is usually a good idea, as it will save processing time and keep character data as character.

Corrected:

Again, setting this to FALSE is usually a good idea, as it will save processing time and keep character data as character.

Section 15.2 on Page 193

GGally has been updated so that params argument in ggpairs is deprecated and its use will cause an error as below.

> data(economics, package='ggplot2')
> GGally::ggpairs(economics[, c(2, 4:6)], params=list(labelSize=8))
Error in stop_if_params_exist(params): 'params' is a deprecated argument.  Please 'wrap' the function to supply arguments. help("wrap", package = "GGally")

The fix is to call ggpairs without the params argument and use the ggplot2 function theme with axis.text set to ggplot2::element_text(size=2) to set the label text size.

> GGally::ggpairs(economics[, c(2, 4:6)]) + ggplot2::theme(axis.text=ggplot2::element_text(size=2))

Section 19.2 on Page 293

Depending on the version of R and the arm package being used, using bayesglm with the subset argument can lead to an error as follows:

> resultsB <- vector(mode="list", length=length(theYears))
> names(resultsB) <- theYears
> for(i in theYears)
+ {
+     # fit model with Cauchy priors with a scale of 2.5
+     resultsB[[as.character(i)]] <- 
+         arm::bayesglm(Vote ~ Race + Income + Gender + Education, 
+                       data=ideo, subset=Year==i, 
+                       family=binomial(link="logit"), 
+                       prior.scale=2.5, prior.df=1)
+ }

Error in lm.fit(x = as.matrix(x[good.star, ]) * w.star, y = z.star * w.star) : 
  incompatible dimensions
In addition: Warning message:
In as.matrix(x[good.star, ]) * w.star :
  longer object length is not a multiple of shorter object length

Not all users will experience this error as it is dependent on the version number of both R and arm. A simple workaround is to remove the subset argument and to subset the data using square bracket ([, ]) notation:

> resultsB <- vector(mode="list", length=length(theYears))
> names(resultsB) <- theYears
> for(i in theYears)
+ {
+     # fit model with Cauchy priors with a scale of 2.5
+     resultsB[[as.character(i)]] <- 
+         arm::bayesglm(Vote ~ Race + Income + Gender + Education, 
+                       data=ideo[ideo$Year == i, ], 
+                       family=binomial(link="logit"), 
+                       prior.scale=2.5, prior.df=1)
+ }

Section 21.3 on Page 328

There are two contradictory blocks of code to load the AT&T ticker data:

> require(quantmod)
> load("data/att.rdata")

and

> require(quantmod)
> att <- getSymbols("T", auto.assign=FALSE)

The block containing load is a misprint and should be ignored. To correctly obtain the data run the block that uses getSymbols to download the data from the Internet:

> require(quantmod)
> att <- getSymbols("T", auto.assign=FALSE)

Section 22.2 on Page 347

For reasons unknown the World Bank has removed the zip file of the world map shapefile and its new location is unclear. In the meantime, a copy of the zip file has been posted at http://jaredlander.com/data/worldmap.zip.

The original code

> download.file(url="http://maps.worldbank.org/overlays/2712.zip", destfile="data/worldmap.zip", method="curl")

should be replace with:

> download.file(url="http://jaredlander.com/data/worldmap.zip", destfile="data/worldmap.zip", method="curl")

Section 24.2.2 on Page 377

In copies from the first printing the wrong function is output when running the print command without parenthesis. The editions with error look like:

> print
standardGeneric for "print" defined from package "base"

function (x, ...) 
standardGeneric("print")
<environment: 0x000000000c3dcdc0>
Methods may be defined for arguments: x
Use  showMethods("print")  for currently available ones.

That is for an S4 version of print. The intent was to show the S3 version as follows:

> print
function (x, ...) 
UseMethod("print")
<bytecode: 0x00000000081e2bb0>
<environment: namespace:base>

Appendix A.7 on Page 394

The book The Elements of Statistical Learning is erroneously attributed to Hastie, Tibshirani and Hastie when it should be Hastie, Tibshirani and Friedman.

Index of People on Page 433

In the first printing the index incorrectly lists Trevor Tibshirani rather than Robert Tibshirani.