Embed plots into Rmarkdown

[adamaki]

Hi @bradduthie, apologies for missing your Rmarkdown session last week. I've been reading through your notes and having a play with it this morning. However, I'm having trouble inserting plots that I've created previously.

So I have a script that creates a ggplot from a dataset and saves the ggplot object in a variable, e.g. 'wdflp' and also creates a panel of six ggplots using the cowplot package saved in another variable, e.g. 'fig7'. How do I insert these objects into my Rmarkdown script? Everything I've tried comes up with an error when I try to knit it, e.g. 'fig7 not found'. All of the objects are in my global environment.

I'm sure there must be a simple solution, but no amount of scouring the web has provided an answer so far.

[bradduthie]

Hi @adamaki, and no worries! Do you have a sample of the code, just to see what's going on? Is the challenge in getting the relevant variables read into the Rmd, or producing the figure itself? E.g., can you tell Rmarkdown to print off the variable successfully with something the below:

print(variable_in_Rmd);

I suspect it's a problem of reading in fig7, given your error message?

[adamaki]

Sure, so fig7 is in the global environment as a 'List of 9' produced using cowplot. I've tried several variations in Rmd but currently the code is based on your suggestion:

```{r echo=FALSE}
print(fig7)
```
Using the print(fig7) code, I can get Rmd to show the plot when I click the green play button to the right of the line of code, but I get the same error when I run it.

[bradduthie]

I think I see what you mean. You're saying that you can run the line of code in the console and get the plot, but when you attempt to knit the Rmarkdown document, it does not work? If so, the problem might be that you have fig7 read into the console, but not into the Rmarkdown document itself. For example, assume you define some variable rand_x in the console as below.

rand_x <- rnorm(n = 100);

The 100 random normal values of rand_x can be printed in the R console now, but Rmarkdown won't recognise them unless you also have the above line of code in the Rmarkdown file itself. So in the case of fig7, the data and analyses underlying it needs all be run inside Rmarkdown (i.e., the entire script that produces the result you want in the console needs to be recreated with the Rmd). You could also save the variables to an R data file, and then read them into the Rmarkdown script, if the variables are quite large and you don't want to reproduce them every time you knit.

I hope I've understood your question correctly? Apologies if I've missed something.

[adamaki]

Ah I was hoping to avoid that! The figure needs 200 lines of code to draw it so I wanted to avoid putting all the code into my Rmd so it wouldn't get messy. I was hoping Rmd would be smart enough to be able to drop in an object created in a different script and knit it without any problems.

I tried saving fig7 as an R data file and that's done the job. I simply called readRDS('fig7.rds') and it's read it into my Rmd knit. However, the plots are all squished and the text is out of proportion, which I guess is a formatting issue and another problem that I'll need to work through...

Thanks for your help!

[bradduthie]

Glad it worked! Yeah, the formatting is another issue -- I've not found a better way to do it than just trial and error until things come out looking they way that I want them.

[bradduthie]

Oh, hey @adamaki -- do you have any suggestions of where it might be useful in the notes to make the point that objects in the console cannot be accessed by Rmarkdown?

[adamaki]

@bradduthie I guess you could add a note in line 439 after the plot...

[bradduthie]

@adamaki Awesome, thanks!