Print ggplot2 and lm objects in a loop in RMarkdown

Question

This is a follow-up question to this one.

In the aforementioned question I could nicely implement a heading creating and the printing of a ggplot2 object in a loop.

Now I have a new problem: I need to print the summary of a model in the loop too. The problem is that it doesn't work if I have the asis option.

Any ideas?

[Reprex:]

---
title: "Untitled"
output:
  html_document:
    theme: united
    highlight: tango
    toc: true
    toc_float:
      collapsed: false
      smooth_scroll: false
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = FALSE, message = FALSE, warning = FALSE)
```

```{r}
library(dplyr)
library(ggplot2)

df <- datasets::iris %>% 
  dplyr::as_tibble()
```

```{r species_loop, results='asis'}
for(i in c("setosa", "versicolor", "virginica")) {

  cat(paste0("\n\n## ", i, "\n"))

  df_filtered <- df %>% 
    dplyr::filter(Species == i)

  p <-  df_filtered %>% 
    ggplot2::ggplot(ggplot2::aes(Sepal.Length, Petal.Length)) + 
    ggplot2::geom_point()
  print(p)

  my_model <- lm(Sepal.Length ~ Petal.Length, data = df_filtered)

  summary(my_model) %>% 
    print()
}
```

## I need the printing output of the model to look like this:

```{r}
df_filtered <- df %>% 
    dplyr::filter(Species == "setosa")

my_model <- lm(Sepal.Length ~ Petal.Length, data = df_filtered)

  summary(my_model) %>% 
    print()
```

I'm not sure exactly what you mean because it being a "part 2" question, but when I replace print() with ```broom::glance(my_model)```, you can create a tidy table of your result and pipe it into a kable() — Corey Pembleton, Oct 01 '19 at 18:00
Yeah I know, but in my case I can't tidy it.. I need to use the print method of the object — FMM, Oct 01 '19 at 18:02
The discussion in [this issue](https://github.com/yihui/knitr/issues/1137), while fairly old, seems to indicate the asis plus verbatim text output is going to be difficult. — aosmith, Oct 01 '19 at 18:42

score 1 · Accepted Answer · answered Oct 01 '19 at 21:11

It looks like this is a case for a template document and knitr::knit_expand(). A canonical answer is here. Working with strings sent me to this answer, as well, since I wasn't entire sure how the "{{" notation worked in this approach.

This approach involves making a template R markdown document in the same directory. Depending on whether you want to include code chunks or need chunk names, that could look something like:

## {{species}}

```{r species_loop-{{species}}, echo = FALSE}
df_filtered <- df %>% 
     dplyr::filter(Species == "{{species}}")

df_filtered %>% 
     ggplot2::ggplot(ggplot2::aes(Sepal.Length, Petal.Length)) + 
     ggplot2::geom_point()

my_model <- lm(Sepal.Length ~ Petal.Length, data = df_filtered)
summary(my_model)
```

Then your master document code could look like:

```{r}
library(dplyr)
library(ggplot2)

df <- datasets::iris %>% 
  dplyr::as_tibble()
```

```{r, include = FALSE}
src <- lapply(c("setosa", "versicolor", "virginica"),
             function(species) {
                  knitr::knit_expand(file = "template.Rmd")
             }
)
```

`r knitr::knit(text = unlist(src))`

score 0 · Answer 2 · answered Oct 01 '19 at 18:18

0

When I replace the summary() line with a tidymodel, you can pipe into a kable:

 print(broom::tidy(my_model) %>% knitr::kable())

You'll have to adjust the settings of the tidy() call to include more outputs, but its working for me:

full code:

for(i in c("setosa", "versicolor", "virginica")) {

  cat(paste0("\n\n## ", i, "\n"))

  df_filtered <- df %>% 
    dplyr::filter(Species == i)

  p <-  df_filtered %>% 
    ggplot2::ggplot(ggplot2::aes(Sepal.Length, Petal.Length)) + 
    ggplot2::geom_point()
  print(p)

  my_model <- lm(Sepal.Length ~ Petal.Length, data = df_filtered)

 print(broom::tidy(my_model) %>% knitr::kable())


}

answered Oct 01 '19 at 18:18

Corey Pembleton

717
9
23

Thanks for your answer but in my real case the model doesn't work with tidy(). I need to use the object's print method. It's like having two chunks inside the loop: one with 'asis' and one without it – FMM Oct 01 '19 at 18:20
Okay cool, wasn't sure if it was something else. A shame because it seems like the simplest way to include the models formatted! Good Luck – Corey Pembleton Oct 01 '19 at 18:22

Print ggplot2 and lm objects in a loop in RMarkdown

2 Answers2