Control ggplot2 facet height independently from number of row facets

Question

Problem

ggplot2/knitr forces plots (in LaTeX) to have the default aspect ratio. Two Plots which have different number of facets result in plots with facet heights being different. This is not nice.

I was looking for a solution which controls the ggplot2 facet height independently from the number of row facets.

Edited Sept. 9th: The solution should work within one chunk as both plots are in one chunk (as in the example code). So tweaking knitr's figure related chunk options is not viable, as the chunk option applies to both plots equally.

Solution - but having problems

Putting each plot into a gridExtra box and scaling it's height with the number of facets, results in facet heights which are no longer different: nicer but not perfect.

• Plots appear on the page with too much margin, even there's remaining text below so plots shall move more closer.
• Seems plots manipulated by this get cropped.

Is there a smarter and problem-free solution which controls the ggplot2 facet height independently from the number of row facets?

\documentclass[a4paper]{article}
\usepackage[margin=1in]{geometry}

\begin{document}

<<setup, results='asis', message=FALSE, echo=FALSE>>=
require(ggplot2)
require(gridExtra)

### Generate two data frames

# Data frame with 2 classes
db.small = data.frame (
  class = as.factor(c(rep("A", 12), rep("B", 12))),
  month = as.factor(rep(1:12, 2)),
  value = runif(2*12, 0, 100)
  )

# Data frame with 5 classes
db.large = data.frame (
  class = as.factor(c(rep("A", 12), rep("B", 12), rep("C", 12), rep("C", 12), rep("D", 12))),
  month = as.factor(rep(1:12, 5)),
  value = runif(5*12, 0, 100)
)

# Generate plots
plot1 = ggplot(db.small, aes(month, value)) + geom_bar(stat="identity") + facet_grid(class ~ .) + ylim(c(0, 100))
plot2 = ggplot(db.large, aes(month, value)) + geom_bar(stat="identity") + facet_grid(class ~ .) + ylim(c(0, 100))
@

\section{Before: Native plots without modification}
Ggplot2/knitr forces both plots to have the default aspect ratio. As both plots have different number of facets the result is that the facet heights are different. This is not nice.

<<before, results='asis', message=FALSE, echo=FALSE, out.width="0.6\\linewidth">>=
print(plot1)
print(plot2)
@

\pagebreak
\section{After: Plots with modification}
Putting each plot into a gridExtra box and scaling it's height with the number of facets, results in facet heights which are no longer different: nicer but not perfect, see problems below.

<<After, results='asis', message=FALSE, echo=FALSE, out.width="0.6\\linewidth">>=
# Calculate number of facets for the two plots
db.small.numberfacets = length(unique(db.small$class))
db.large.numberfacets = length(unique(db.large$class))

# Define heights for modification of plots
facet.height = 4
other.height = 2

plot1 = plot1 + theme( plot.margin = unit(c(0, 0, 0, 0), "cm"))
plot2 = plot2 + theme( plot.margin = unit(c(0, 0, 0, 0), "cm"))

# Put plots inside a gridExtra box and change height plots
grid.arrange(arrangeGrob(plot1, heights=unit(db.small.numberfacets*facet.height+other.height, "cm")), nrow=1)
grid.arrange(arrangeGrob(plot2, heights=unit(db.large.numberfacets*facet.height+other.height, "cm")), nrow=1)
@


2 Problems:
\begin{itemize}
\item Plots appear on the page with too much margin, even there's remaining text below so plots shall move more closer.
\item Seems the second plot is cropped, see the x-axis label.
\end{itemize}

Lorem ipsum dolor sit amet, consectetuer adipiscing elit, sed diam nonummy nibh euismod tincidunt ut laoreet dolore magna aliquam erat volutpat. Ut wisi enim ad minim veniam, quis nostrud exerci tation ullamcorper suscipit lobortis nisl ut aliquip ex ea commodo consequat.

\end{document}

pdf Output:

Answer 1

Do you know that knitr chunk options can take arbitrary R expressions? The example 40 shows how you can calculate fig.width and fig.height dynamically.

For your case, you can assign a value to fig.height according to the number of levels in the factor variable for faceting. For example,

\documentclass{article}

\begin{document}
<<setup>>=
library(ggplot2)
calc_height = function(f) length(levels(f))
p = ggplot(diamonds, aes(color)) + geom_bar()
@

<<test-a, fig.width=7, fig.height=calc_height(diamonds$cut)>>==
p + facet_grid(cut ~ .)
@

<<test-b, fig.width=7, fig.height=calc_height(diamonds$clarity)>>==
p + facet_grid(clarity ~ .)
@

\end{document}

Output: