A quick one today. If you work with economic data, you’ll be confronted to NACE code sooner or later. NACE stands for Nomenclature statistique des Activités économiques dans la Communauté Européenne. It’s a standard classification of economic activities. It has 4 levels, and you can learn more about it here.

Each level adds more details; consider this example:

C - Manufacturing
C10 - Manufacture of food products
C10.1 - Processing and preserving of meat and production of meat products
C10.1.1 - Processing and preserving of meat
C10.1.2 - Processing and preserving of poultry meat
C10.1.3 - Production of meat and poultry meat products

So a company producing meat and poultry meat products would have NACE code level 4 C10.1.3 with it. Today for work I had to create a nice visualisation of the hierarchy of the NACE classification. It took me a bit of time to find a nice solution, so that’s why I’m posting it here. Who knows, it might be useful for other people. First let’s get the data. Because finding it is not necessarily very easy if you’re not used to navigating Eurostat’s website, I’ve put the CSV into a gist:

library(tidyverse)
library(data.tree)
library(igraph)
library(GGally)
nace_code <- read_csv("https://gist.githubusercontent.com/b-rodrigues/4218d6daa8275acce80ebef6377953fe/raw/99bb5bc547670f38569c2990d2acada65bb744b3/nace_rev2.csv")
## Parsed with column specification:
## cols(
##   Order = col_double(),
##   Level = col_double(),
##   Code = col_character(),
##   Parent = col_character(),
##   Description = col_character(),
##   `This item includes` = col_character(),
##   `This item also includes` = col_character(),
##   Rulings = col_character(),
##   `This item excludes` = col_character(),
##   `Reference to ISIC Rev. 4` = col_character()
## )
head(nace_code)
## # A tibble: 6 x 10
##    Order Level Code  Parent Description `This item incl… `This item also…
##    <dbl> <dbl> <chr> <chr>  <chr>       <chr>            <chr>           
## 1 398481     1 A     <NA>   AGRICULTUR… "This section i… <NA>            
## 2 398482     2 01    A      Crop and a… "This division … This division a…
## 3 398483     3 01.1  01     Growing of… "This group inc… <NA>            
## 4 398484     4 01.11 01.1   Growing of… "This class inc… <NA>            
## 5 398485     4 01.12 01.1   Growing of… "This class inc… <NA>            
## 6 398486     4 01.13 01.1   Growing of… "This class inc… <NA>            
## # … with 3 more variables: Rulings <chr>, `This item excludes` <chr>,
## #   `Reference to ISIC Rev. 4` <chr>

So there’s a bunch of columns we don’t need, so we’re going to ignore them. What I’ll be doing is transforming this data frame into a data tree, using the {data.tree} package. For this, I need columns that provide the hierarchy. I’m doing this with the next chunk of code. I won’t explain each step, but the idea is quite simple. I’m using the Level column to create new columns called Level1, Level2, etc. I’m then doing some cleaning:

nace_code <- nace_code %>%
  select(Level, Code)

nace_code <- nace_code %>%
  mutate(Level1 = ifelse(Level == 1, Code, NA)) %>%
  fill(Level1, .direction = "down") %>%  
  mutate(Level2 = ifelse(Level == 2, Code, NA)) %>%
  fill(Level2, .direction = "down") %>%  
  mutate(Level3 = ifelse(Level == 3, Code, NA)) %>%
  fill(Level3, .direction = "down") %>%  
  mutate(Level4 = ifelse(Level == 4, Code, NA)) %>%  
  filter(!is.na(Level4))

Let’s take a look at how the data looks now:

head(nace_code)
## # A tibble: 6 x 6
##   Level Code  Level1 Level2 Level3 Level4
##   <dbl> <chr> <chr>  <chr>  <chr>  <chr> 
## 1     4 01.11 A      01     01.1   01.11 
## 2     4 01.12 A      01     01.1   01.12 
## 3     4 01.13 A      01     01.1   01.13 
## 4     4 01.14 A      01     01.1   01.14 
## 5     4 01.15 A      01     01.1   01.15 
## 6     4 01.16 A      01     01.1   01.16

I can now create the hierarchy using by creating a column called pathString and passing that data frame to data.tree::as.Node(). Because some sections, like C (manufacturing) are very large, I do this separately for each section by using the group_by()–nest() trick. This way, I can create a data.tree object for each section. Finally, to create the plots, I use igraph::as.igraph() and pass this to GGally::ggnet2(), which takes care of creating the plots. This took me quite some time to figure out, but the result is a nice looking PDF that the colleagues can now use:

nace_code2 <- nace_code %>%
  group_by(Level1, Level2) %>%
  nest() %>%
  mutate(nace = map(data, ~mutate(., pathString = paste("NACE2",
                                       Level1,
                                       Level2,
                                       Level3,
                                       Level4,
                                       sep = "/")))) %>%
  mutate(plots = map(nace, ~as.igraph(as.Node(.)))) %>%
  mutate(plots = map(plots, ggnet2, label = TRUE))


pdf("nace_maps.pdf")
pull(nace_code2, plots)
dev.off()

Here’s how the pdf looks like:

If you want to read more about {data.tree}, you can do so here and you can also read more about the ggnet2() here.

Hope you enjoyed! If you found this blog post useful, you might want to follow me on twitter for blog post updates and buy me an espresso or paypal.me, or buy my ebook on Leanpub.

Buy me an Espresso