plot with ggplot in for-loop doesn´t compile to PDF

Question

I created a report using knitR (.Rnw) to compile it as a PDF. As I would like to plot the same figure for every single questions I created a plot in a for-loop. Unfortunately I get a warning in the PDF and don´t realy get what causes the error...

Here is the chunk where I believe the problem starts: the loop runs fine within R but does not compile as a PDF (see the whole code below). I tried various labels, print functions and other stuff but couldn´t find a solution.

<<echo=FALSE, warning=T, message=F>>=

for(i in 1:3){
  cat(paste("\\subsection{",titel[i],"}\n", sep=""))
  cat(paste("Figure \\ref{class",i,"} \n", sep=""))
   cat(paste("\\begin{figure}[H] \n", sep=""))
   cat(paste("\\begin{center} \n", sep=""))
     cat(paste("\\includegraphics[width=1\\textwidth,", 
            "height=.47\\textheight,keepaspectratio]{class",i,".pdf}\\caption{",titel[i],"}\n", sep=""))

    cat(paste("\\label{class",i,"}" \n, sep=""))
    cat(paste("\\end{center} \n",sep=""))
  cat(paste("\\end{figure} \n",sep=""))



  p <- ggplot(data[!is.na(data$F17),], aes_string(x=Fragen[i], y="..prop..", group = "1", fill="F17"))+ 
       geom_bar()+
       facet_grid(F17~.)+
       geom_text(aes(label = scales::percent(..prop..),
              y= ..prop.. ), stat= "count", vjust = -.5, size=3) + 
       ylab("Prozent")+
       xlab(titel[i])+
       scale_fill_manual(name="Individuals", values=colorScheme)#+
       #theme_mine


 pdfnam<-paste("class",i,".pdf",sep="") #produce a plot for each class
 pdf(file=pdfnam,width=12, height = 4)
 #gridExtra::grid.arrange(p, q)
 print(p)
 dev.off()
}

@

Here is the whole code for replication:

\documentclass{article}

\usepackage{amsmath,amssymb,amstext}
\usepackage{graphicx}
\usepackage{geometry}
\geometry{top=15mm, left=25mm, right=25mm, bottom=25mm,headsep=10mm,footskip=10mm} 
\usepackage{xcolor}
\usepackage{float}
\usepackage[T1]{fontenc} % Umlaute
\usepackage[utf8]{inputenc}
\inputencoding{latin1}

\begin{document}
\parindent 0pt

\title{title} 
\maketitle

<<echo=FALSE, warning=FALSE, message=FALSE>>=
library(ggplot2)
library(reshape)
library(knitr)
library(doBy)
library(dplyr)


opts_chunk$set(fig.path='figure/graphic-', fig.align='center', fig.show='hold',fig.pos='!ht',
           echo=FALSE,warning = FALSE) 

@

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

 # data and other useful stuff

 data <- data.frame(F1 = c("A", "A", "B", "C"), # answers to question 1, ...
                    F2 = c("A", "B", "B", "C"),
                    F3 = c("A", "B", "C", "C"),
                    F17 = c("K", "L", "L", "M")) # K, L and M are a certain individual. L answered twice.

 # colour scheme:
 GH="#0085CA"; H="#DA291C"; BV="#44697D"
 colorScheme <- c(BV, H, GH)

 # individual theme for plots:
theme_mine = theme(plot.background = element_rect(fill = "white"),
               panel.background = element_rect(fill = "white", colour = "grey50"),
               text=element_text(size=10, family="Trebuchet MS"))

# a vector with the variable names from "data" (F1, F2, F3).
Fragen <- c(paste0('F',seq(1:3), sep=""))

# question title for labeling the plots:
titel <- c("Q1", "Q2", "Q3", "Q17")

@


\begin{figure}[h]
\begin{center}
<<echo=FALSE, fig.width=9.6, fig.height=6, warning=FALSE>>=

p <- ggplot(data, aes(x=F17))+
     geom_bar(fill = colorScheme)+
     xlab(titel[4])+
     #geom_text(aes(label = scales::percent(..prop..),
     #          y= ..prop.. ), stat= "count", vjust = -.5, size=3) + 
     ylab("Absolut")+
     theme_bw()
     #theme_mine   # does not work properly yet.
p

@
\caption{figa}
\label{figa}
\end{center}
\end{figure}

\section{individual plots}

<<echo=FALSE, warning=T, message=F>>=

# here is where the problem starts: the loop runs fine within R but does not compile to an PDF.

for(i in 1:3){
  cat(paste("\\subsection{",titel[i],"}\n", sep=""))
  cat(paste("Figure \\ref{class",i,"} \n", sep=""))
   cat(paste("\\begin{figure}[H] \n", sep=""))
   cat(paste("\\begin{center} \n", sep=""))
     cat(paste("\\includegraphics[width=1\\textwidth,", 
            "height=.47\\textheight,keepaspectratio]{class",i,".pdf}\\caption{",titel[i],"}\n", sep=""))

    cat(paste("\\label{class",i,"}" \n, sep=""))
    cat(paste("\\end{center} \n",sep=""))
  cat(paste("\\end{figure} \n",sep=""))



  p <- ggplot(data[!is.na(data$F17),], aes_string(x=Fragen[i], y="..prop..", group = "1", fill="F17"))+ 
       geom_bar()+
       facet_grid(F17~.)+
       geom_text(aes(label = scales::percent(..prop..),
              y= ..prop.. ), stat= "count", vjust = -.5, size=3) + 
       ylab("Prozent")+
       xlab(titel[i])+
       scale_fill_manual(name="Individuals", values=colorScheme)#+
       #theme_mine


 pdfnam<-paste("class",i,".pdf",sep="") #produce a plot for each class
 pdf(file=pdfnam,width=12, height = 4)
 #gridExtra::grid.arrange(p, q)
 print(p)
 dev.off()
}

@


\end{document}

Thanks a lot in advance!

Answer 1

There are several advices I'm trying to convey using your reproducible script. Thanks for providing it, it helps a lot.

Usually you can use knitr to produce the graphs provided you have a plot or print(ggplot_object)in your R chunk. In your example you try to mix up a begin{Figure} with R code to produce your plot object. You don't need to use it. Knitr will provide the tools to create a full plot objet, pointing to the figure path (default) which is located in the same folder as your .Rnw script. I'm giving you an example on how to do that below.

The only drawback is that if you try to read your tex file, it not rendered as nicely as if you had created that code yourself (what I mean is that when you edit your tex file, everything is in one row, but this is not a criticism of knitr which is just so good). The other choice, which you have tried too, is to save the figure somewhere in one folder, and then load it with a tex command. In the example below, I'm using your script to give you an example how to include a figure this way. I hope that will be usefull to you.

\documentclass{article}

\usepackage{amsmath,amssymb,amstext}
\usepackage{graphicx}
\usepackage{geometry}
\geometry{top=15mm, left=25mm, right=25mm, bottom=25mm,headsep=10mm,footskip=10mm} 
\usepackage{xcolor}
\usepackage{float}
\usepackage[T1]{fontenc} % Umlaute
\usepackage[utf8]{inputenc}
\inputencoding{latin1}
\usepackage{hyperref}
\begin{document}
\parindent 0pt

\title{title} 
\maketitle

<<echo=FALSE, warning=FALSE, message=FALSE>>=
library(ggplot2)
library(reshape)
library(knitr)
library(doBy)
library(dplyr)


opts_chunk$set(fig.path='figure/graphic-', fig.align='center', fig.show='hold',fig.pos='!ht',
           echo=FALSE,warning = FALSE) 

@

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

 # data and other useful stuff

 data <- data.frame(F1 = c("A", "A", "B", "C"), # answers to question 1, ...
                    F2 = c("A", "B", "B", "C"),
                    F3 = c("A", "B", "C", "C"),
                    F17 = c("K", "L", "L", "M")) # K, L and M are a certain individual. L answered twice.

 # colour scheme:
 GH="#0085CA"; H="#DA291C"; BV="#44697D"
 colorScheme <- c(BV, H, GH)

 # individual theme for plots:
theme_mine = theme(plot.background = element_rect(fill = "white"),
               panel.background = element_rect(fill = "white", colour = "grey50"),
               text=element_text(size=10, family="Trebuchet MS"))

# a vector with the variable names from "data" (F1, F2, F3).
Fragen <- c(paste0('F',seq(1:3), sep=""))

# question title for labeling the plots:
titel <- c("Q1", "Q2", "Q3", "Q17")

@

First chunk uses the knitr output to place the figures, if you use ggplot don't
forget to print your plot : \textbf{print(p)} otherwise it won't work . All your
arguments are passed through chunk options. So where you tried to have them in the text,
they are simply placed as other options to your chunk (see below). I have used
the following options to reproduce your example.
\begin{itemize}
\item fig.width=9.6
\item fig.height=6
\item fig.pos='h',
\item fig.cap="figa"
\item fig.lp="figa"
\item fig.align='center'
\end{itemize}

<<echo=FALSE, fig.width=9.6, fig.height=6, warning=FALSE, fig.pos='h', fig.cap="figa",fig.lp="figa", fig.align='center'>>=

p <- ggplot(data, aes(x=F17))+
     geom_bar(fill = colorScheme)+
     xlab(titel[4])+
     #geom_text(aes(label = scales::percent(..prop..),
     #          y= ..prop.. ), stat= "count", vjust = -.5, size=3) + 
     ylab("Absolut")+
     theme_bw()
     #theme_mine   # does not work properly yet.
print(p)

@




\section{individual plots}

For individual plots we will use your script to generate the figure environment.
To produce latex you need to pass the option 'asis'.

<<generate_latex,echo=FALSE, warning=T, message=F, results='asis'>>=
for(i in 1:3){
  cat(paste("\\subsection{",titel[i],"}\n", sep=""))
  cat(paste("Figure \\ref{class",i,"} \n", sep=""))
   cat(paste("\\begin{figure}[H] \n", sep=""))
   cat(paste("\\begin{center} \n", sep=""))
     cat(paste("\\includegraphics[width=1\\textwidth,", 
            "height=.47\\textheight,keepaspectratio]{class",i,".pdf}\\caption{",titel[i],"}\n", sep=""))
    cat(paste("\\label{class",i,"} \n", sep=""))
    cat(paste("\\end{center} \n",sep=""))
  cat(paste("\\end{figure} \n",sep=""))
}
@
Now we need to save those figures.  By default in knitr figures are saved in the \textit{figure}
subfolder and path is set to \textit{figure/myfigure} in the includegraphics
command in the tex file.

<<plot,echo=FALSE, warning=T, message=F, fig.keep='all',fig.show='hide', results='hide'>>= 
for(i in 1:3){
  p <- ggplot(data[!is.na(data$F17),], aes_string(x=Fragen[i], y="..prop..", group = "1", fill="F17"))+ 
       geom_bar()+
       facet_grid(F17~.)+
       geom_text(aes(label = scales::percent(..prop..),
              y= ..prop.. ), stat= "count", vjust = -.5, size=3) + 
       ylab("Prozent")+
       xlab(titel[i])+
       scale_fill_manual(name="Individuals", values=colorScheme)#+
       #theme_mine
  print(p)
}
@
Now the other way to do it, as in sweave is just to save the plot where you
want, this is the old sweave way, which I tend to use, I gave some explanation
on how to arrange folders
\href{https://stackoverflow.com/questions/46920038/how-to-get-figure-floated-surrounded-by-text-etc-in-r-markdown/46962362#46962362}{example script}
In sweave you save those files using pdf() or png() or
whatever device and set the graphics path to those figures using
\textit{\\graphicspath{{/figure}}} in the preamble. If you want to set your
graphics path to another folder you can set path using command like
\textit{\\graphicspath{{../../figure}}}. This will set your path to the
grandparent folder.

Here I'm creating a directory, if existing, the code will still proceed with no
warnings : \textit{dir.create(imagedir, showWarnings = FALSE)}.

<<plot_manual,echo=FALSE, warning=T, message=F,results='hide'>>=
imagedir<-file.path(getwd(), "figure")
dir.create(imagedir, showWarnings = FALSE) # use show warning = FALSE
pdfnam<-paste0(imagedir,"/class4.pdf") #produce a plot for each class
pdf(file=pdfnam,width=8, height = 4)
for(i in 4){
  p <- ggplot(data[!is.na(data$F17),], aes_string(x=Fragen[i], y="..prop..", group = "1", fill="F17"))+ 
      geom_bar()+
      facet_grid(F17~.)+
      geom_text(aes(label = scales::percent(..prop..),
              y= ..prop.. ), stat= "count", vjust = -.5, size=3) + 
      ggtitle("manually saved plot")+
      ylab("Prozent")+
      xlab(titel[i])+
      scale_fill_manual(name="Individuals", values=colorScheme)
  print(p)
}
 dev.off()
@
\subsection{section 4}
This is Figure \ref{class4} 
\begin{figure}[H] 
 \begin{center} 
   \includegraphics[width=1\textwidth,height=.47\textheight,keepaspectratio]{figure/class4.pdf}
   \caption{Manually edited caption for figure 4}
   \label{class4} 
  \end{center}
\end{figure} 

\end{document}