How can I make a variable-length report using Rmarkdown? - r

Question

I am using R / Rmarkdown / knitr to generate multiple reports (pdfs) via render(), but the content / length of the reports will vary depending on certain characteristics of the underlying data.
As an example, let's say I have 10 different datasets of 50 variables each and I'm examining a correlation matrix of all 50 variables in the data. I want to produce a report for each dataset that has a new page for each variable pair that has a correlation that is greater than 0.5 and each variable pair that has a correlation that is less than -0.5. The number of correlations that will meet these thresholds will vary by dataset, and thus the report length / number of pages will vary by dataset.
I've learned to use {asis, echo = somecondition, eval = somecondition} to evaluate whether an entire section needs to be included (e.g., when there are no negative correlations less than -0.5). I have also seen solutions utilizing 'for' loops when there might be variable-length arguments across reports, but those solutions don't include printing each result on a new page. I'd also like to include section headers on each of the pages reporting the correlations as well.
The difficulty for me is that any solution I can think of requires nesting chunks of text and r code within one another. For some sample Rmd code of how I am approaching the problem, I've tried to print a new histogram for each small dataset on a new page, using "```" to denote where three ticks would usually be as to not mess up the sample code formatting:
"```"{r, echo = FALSE}
datlist <- list(df1 = rnorm(100), df2 = rnorm(100), df3 = rnorm(100)) # fake data
"```"
Some Text Introducing the Report
"```"{'asis', eval = length(datlist) > 0} # evaluating if the section is to be included
"```"{r, echo = FALSE, eval = length(datlist) > 0}
for(i in 1:length(datlist)){ # starting the variable-length scope
"```"{'asis', eval = length(datlist) > 0} # the information to be included on each new page
\newpage
\section{`r (names(datlist[i]))`}
Here is a histogram of the data found in `r (names(datlist[i]))`.
`r hist(unlist(datlist[i]))`
"```"
} # closing the for loop above
"```"
"```"
Any help, including a solution using a completely different approach, is most welcome.

Answer 1

A correlation is always between two variables so I am unsure wether this is what you want, but the following code will display the correlation of all pairs of variables that are greater than 0.5 in absolute value.
---
title: "Untitled"
author: "Author"
date: "18 November 2019"
output: pdf_document
---
```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE)
```
```{r, echo = FALSE}
datlist <- data.frame(var1 = rnorm(100),
var2 = rnorm(100),
var3 = rnorm(100)) # fake data
# add some correlations
datlist$var4 <- datlist$var1*(rnorm(100,0,0.05)+1)
datlist$var5 <- datlist$var3*(rnorm(100,0,0.05)-1)
# get all correlations, there is probably an easier way of doing this...
corlist <- as.data.frame(t(combn(colnames(datlist),2)))
corlist$cor <- apply(corlist,1,function(x) {
cor(datlist[,x[1]],datlist[,x[2]])
})
```
Some Text Introducing the Report
```{r, results='asis', echo=F}
apply(corlist[abs(corlist$cor)>0.5,],1, function(x) {
cat('\n')
cat("# Correlation between ", x[1], " and ",x[2],"\n")
cat("The correlation between both variables was ", x[3], ".\n")
})
```
Of course you can extend the content of the loop to do whatever you want with the variables.
Original solution from here