Hiding NA's when printing a dataframe in knitr

Question

I'm trying to print a table in knitr from a data frame using xtable . The table in the example below has the dimensions 3x7 but the third row only has one value, in the second column. The rest of the cells in the third row are 'NA'.

When I compile the document, is there a way to prevent knitr from printing the NA's in the third row, so instead of NA I just have blank space?

It feels like this should be a simple solution but I can't work out where/how to hide the NA's. Is it a change I need to make to the data frame or is it an xtable or knitr option I need to change?

Sample knitr code:

\documentclass{article}

<< data1, echo=FALSE,  warning=FALSE, message=FALSE >>=

require(xtable)

  FY.2014 <- 0.019
  FY.2015 <- ((7000)  - (6925.9)) / (6925.9)
  FY.2016 <- ((8000)  - (7000))   / (7000)
  FY.2017 <- ((9000)  - (8000))   / (8000)
  FY.2018 <- ((10000) - (9000))   / (9000)
  FY.2019 <- ((11000) - (10000))  / (10000)

  PC      <- data.frame(FY.2014, FY.2015, FY.2016, FY.2017, FY.2018, FY.2019)
  PC.1    <- paste(round(PC*100, digits=1), "%", sep="")


 FY.2014 <- 130.1
 FY.2015 <- 7000  - 6925.9
 FY.2016 <- 8000  - 7000
 FY.2017 <- 9000  - 8000
 FY.2018 <- 10000 - 9000
 FY.2019 <- 11000 - 10000

 AB      <- data.frame(FY.2014, FY.2015, FY.2016, FY.2017, FY.2018, FY.2019)
 AB.1    <- paste(round(AB , digits = 2))


    FY.2014 <- as.numeric(c(""))
    FY.2015 <- 7242.9
    FY.2016 <- as.numeric(c(""))
    FY.2017 <- as.numeric(c(""))
    FY.2018 <- as.numeric(c(""))
    FY.2019 <- as.numeric(c(""))

    PF      <- data.frame(FY.2014, FY.2015, FY.2016, FY.2017, FY.2018, FY.2019)
    PF.1    <- paste(round(PF , digits = 2))

     FTable  <- rbind( PC.1, AB.1, PF.1)

      rownames(FTable) <- c( 'Percent Change from the Previous Year', 
                             'Absolute Change from Previous Year', 
                             'December CY13 Forecast')
      colnames(FTable) <- c( 'FY 2014', 'FY 2015', 'FY 2016', 'FY 2017', 'FY 2018',    'FY 2019')

@ 

\begin{document}
<<Table 1 , echo=FALSE, eval=TRUE, results='asis', fig.width = 5, fig.height = 2,     message=FALSE, fig.align='center', warning=FALSE>>=

          xFTable  <- xtable(FTable, big.mark=",")

          print(xFTable) 
@
\end{document}

Solution 1

You can set the knitr option knitr.kable.na = '' '' for blanks, or whatever character you want.

```{r echo=FALSE, results='asis'}
    options(knitr.kable.NA = '')
    knitr::kable(lowerTri, digits=2)
```

Solution 2

The trick I use is a bit brute-force, but it appears to work (in my use-cases, that is):

out <- knitr::kable(...)
cat(gsub('\\bNA\\b', '  ', out), sep='\n')

Comments

[Rich Scriven]

You could try print(xFTable, na.print = "") You may have to use print.default or a different print method to print the doc. But na.print appears in quite a few print methods

[daroczig]

Or if you are happy with markdown tables instead of LaTeX, then you could use pander where you can specify how to print NAs with the missing argument or global option.

[Matt]

Thanks for the suggestion but I tried print(xFTable, na.print="") without any success. I didn't find na.print in the xtable documentation.

[YvanR]

I guess '\\bNA\\n' should be '\\bNA\\b'?

[Costin]

Where do you set it ? It doesn't work for me. Is it knitr.kable.na or knitr.kable.NA ? Do you use simple or double quotes ?

[svannoy]

@Costin, I've added context, I place it in a code chunk.

[Matt]

You need to use knitr version 1.15 or above: github.com/yihui/knitr/releases/tag/v1.15.

[r2evans]

This answer should not longer be relevant given recent changes to knitr. I suggest using @svannoy's answer instead.

[Will]

In my case (Knit to pdf) it turned out to very useful, since the "newer" solution didn't work -- thanks!

[Mikko]

This solution also works for papajas apa_table function (adding the comment to help internet search engines)

[Ryan Hill]

@svannoy is it possible to give a formatted string e.g. 'null' but so it prints in italics?