Econometrics and Free Software by Bruno Rodrigues.

RSS feed for blog post updates.

Follow me on twitter, or check out my Github.

Watch my youtube channel.

Have a question about one of the blog posts? Chat here with me.

Exporting editable plots from R to Powerpoint: making ggplot2 purrr with officer

R

A kind reader let me know that the function create_pptx() is now outdated, and proposed an update which you can find here: here. Thank you @Jeremy!

I was recently confronted to the following problem: creating hundreds of plots that could still be edited by our client. What this meant was that I needed to export the graphs in Excel or Powerpoint or some other such tool that was familiar to the client, and not export the plots directly to pdf or png as I would normally do. I still wanted to use R to do it though, because I could do what I always do to when I need to perform repetitive tasks such as producing hundreds of plots; map over a list of, say, countries, and make one plot per country. This is something I discussed in a previous blog post, Make ggplot2 purrr.

So, after some online seaching, I found the {officer} package. This package allows you to put objects into Microsoft documents. For example, editable plots in a Powerpoint document. This is what I will show in this blog post.

Let’s start by loading the required packages:

library("tidyverse")
library("officer")
library("rvg")

Then, I will use the data from the time use survey, which I discussed in a previous blog post Going from a human readable Excel file to a machine-readable csv with {tidyxl}.

You can download the data here.

Let’s import and prepare it:

time_use <- rio::import("clean_data.csv")


time_use <- time_use %>%
    filter(population %in% c("Male", "Female")) %>%
    filter(activities %in% c("Personal care", "Sleep", "Eating", 
                             "Employment", "Household and family care")) %>%
    group_by(day) %>%
    nest()

I only kept two categories, “Male” and “Female” and 5 activities. Then I grouped by day and nested the data. This is how it looks like:

time_use
## # A tibble: 3 x 2
##   day                         data             
##   <chr>                       <list>           
## 1 Year 2014_Monday til Friday <tibble [10 × 4]>
## 2 Year 2014_Saturday          <tibble [10 × 4]>
## 3 Year 2014_Sunday            <tibble [10 × 4]>

As shown, time_use is a tibble with 2 columns, the first day contains the days, and the second data, is of type list, and each element of these lists are tibbles themselves. Let’s take a look inside one:

time_use$data[1]
## [[1]]
## # A tibble: 10 x 4
##    population activities                time  time_in_minutes
##    <chr>      <chr>                     <chr>           <int>
##  1 Male       Personal care             11:00             660
##  2 Male       Sleep                     08:24             504
##  3 Male       Eating                    01:46             106
##  4 Male       Employment                08:11             491
##  5 Male       Household and family care 01:59             119
##  6 Female     Personal care             11:15             675
##  7 Female     Sleep                     08:27             507
##  8 Female     Eating                    01:48             108
##  9 Female     Employment                06:54             414
## 10 Female     Household and family care 03:49             229

I can now create plots for each of the days with the following code:

my_plots <- time_use %>%
    mutate(plots = map2(.y = day, .x = data, ~ggplot(data = .x) + theme_minimal() +
                       geom_col(aes(y = time_in_minutes, x = activities, fill = population), 
                                position = "dodge") +
                       ggtitle(.y) +
                       ylab("Time in minutes") +
                       xlab("Activities")))

These steps are all detailled in my blog post Make ggplot2 purrr. Let’s take a look at my_plots:

my_plots
## # A tibble: 3 x 3
##   day                         data              plots 
##   <chr>                       <list>            <list>
## 1 Year 2014_Monday til Friday <tibble [10 × 4]> <gg>  
## 2 Year 2014_Saturday          <tibble [10 × 4]> <gg>  
## 3 Year 2014_Sunday            <tibble [10 × 4]> <gg>

The last column, called plots is a list where each element is a plot! We can take a look at one:

my_plots$plots[1]
## [[1]]

Now, this is where I could export these plots as pdfs or pngs. But this is not what I need. I need to export these plots as editable charts for Powerpoint. To do this for one image, I would do the following (as per {officer}’s documentation):

read_pptx() %>%
    add_slide(layout = "Title and Content", master = "Office Theme") %>%
    ph_with_vg(code = print(one_plot), type = "body") %>% 
    print(target = path)

To map this over a list of arguments, I wrote a wrapper:

create_pptx <- function(plot, path){
    if(!file.exists(path)) {
        out <- read_pptx()
    } else {
        out <- read_pptx(path)
    }
    
    out %>%
        add_slide(layout = "Title and Content", master = "Office Theme") %>%
        ph_with_vg(code = print(plot), type = "body") %>% 
        print(target = path)
}

This function takes two arguments, plot and path. plot must be an plot object such as the ones contained inside the plots column of my_plots tibble. path is the path of where I want to save the pptx.

The first lines check if the file exists, if yes, the slides get added to the existing file, if not a new pptx gets created. The rest of the code is very similar to the one from the documentation. Now, to create my pptx I simple need to map over the plots column and provide a path:

map(my_plots$plots, create_pptx, path = "test.pptx")
## Warning in doc_parse_file(con, encoding = encoding, as_html = as_html,
## options = options): Failed to parse QName 'xsi:xmlns:' [202]

## Warning in doc_parse_file(con, encoding = encoding, as_html = as_html,
## options = options): Failed to parse QName 'xsi:xmlns:' [202]

## Warning in doc_parse_file(con, encoding = encoding, as_html = as_html,
## options = options): Failed to parse QName 'xsi:xmlns:' [202]

## Warning in doc_parse_file(con, encoding = encoding, as_html = as_html,
## options = options): Failed to parse QName 'xsi:xmlns:' [202]

## Warning in doc_parse_file(con, encoding = encoding, as_html = as_html,
## options = options): Failed to parse QName 'xsi:xmlns:' [202]

## Warning in doc_parse_file(con, encoding = encoding, as_html = as_html,
## options = options): Failed to parse QName 'xsi:xmlns:' [202]

## Warning in doc_parse_file(con, encoding = encoding, as_html = as_html,
## options = options): Failed to parse QName 'xsi:xmlns:' [202]

## Warning in doc_parse_file(con, encoding = encoding, as_html = as_html,
## options = options): Failed to parse QName 'xsi:xmlns:' [202]

## Warning in doc_parse_file(con, encoding = encoding, as_html = as_html,
## options = options): Failed to parse QName 'xsi:xmlns:' [202]
## [[1]]
## [1] "/home/cbrunos/Documents/b-rodrigues.github.com/content/blog/test.pptx"
## 
## [[2]]
## [1] "/home/cbrunos/Documents/b-rodrigues.github.com/content/blog/test.pptx"
## 
## [[3]]
## [1] "/home/cbrunos/Documents/b-rodrigues.github.com/content/blog/test.pptx"

Here is the end result:

Inside Powerpoint (or in this case Libreoffice), the plots are geometric shapes that can now be edited!

If you found this blog post useful, you might want to follow me on twitter for blog post updates.

Buy me an EspressoBuy me an Espresso