Obtaining package networks

Let’s obtain the three dependency networks defined in rel_packages and store then in a list. I am also calculating three node measures:

• in-degree d_in, the number of edges incident to a node.
• out-degree d_out, the number of edges going out of a node.
• betweenness btw, the number of shortest paths passing through a node.

network_packages <- lapply(rel_packages, function(x){
  
  g <- tbl_graph(edges = ip_table[relation == x], directed = TRUE)
  
  g <- g %>%
    activate(nodes) %>%
    mutate(d_in = centrality_degree(mode = "in"),
         d_out = centrality_degree(mode = "out"),
         btw = centrality_betweenness())
    
  return(g)
})

names(network_packages) <- rel_packages

Let’s examine each of the produced networks.

network_packages

## $Depends
## # A tbl_graph: 232 nodes and 306 edges
## #
## # A directed acyclic simple graph with 1 component
## #
## # Node Data: 232 × 4 (active)
##   name               d_in d_out   btw
##   <chr>             <dbl> <dbl> <dbl>
## 1 R                     0   215     0
## 2 abind                 1     0     0
## 3 arm                   5     0     0
## 4 backports             1     0     0
## 5 BAdatasets            1     0     0
## 6 BAdatasetsSpatial     1     0     0
## # … with 226 more rows
## #
## # Edge Data: 306 × 3
##    from    to relation
##   <int> <int> <chr>   
## 1     1     2 Depends 
## 2     1     3 Depends 
## 3     1     4 Depends 
## # … with 303 more rows
## 
## $Imports
## # A tbl_graph: 299 nodes and 1359 edges
## #
## # A directed acyclic simple graph with 1 component
## #
## # Node Data: 299 × 4 (active)
##   name     d_in d_out     btw
##   <chr>   <dbl> <dbl>   <dbl>
## 1 methods     2    62 191.   
## 2 abind       2     2   0.167
## 3 arm         7     1  18    
## 4 sys         0     3   0    
## 5 askpass     1     3  13    
## 6 tools       0    20   0    
## # … with 293 more rows
## #
## # Edge Data: 1,359 × 3
##    from    to relation
##   <int> <int> <chr>   
## 1     1     2 Imports 
## 2     2     3 Imports 
## 3     4     5 Imports 
## # … with 1,356 more rows
## 
## $Suggests
## # A tbl_graph: 669 nodes and 1987 edges
## #
## # A directed simple graph with 5 components
## #
## # Node Data: 669 × 4 (active)
##   name        d_in d_out     btw
##   <chr>      <dbl> <dbl>   <dbl>
## 1 testthat      11   175 20816. 
## 2 askpass        1     0     0  
## 3 assertthat     2     1     0.5
## 4 BBmisc         3     0     0  
## 5 bit            7     1   294. 
## 6 covr           0   109     0  
## # … with 663 more rows
## #
## # Edge Data: 1,987 × 3
##    from    to relation
##   <int> <int> <chr>   
## 1     1     2 Suggests
## 2     1     3 Suggests
## 3     1     4 Suggests
## # … with 1,984 more rows

We observe that:

• The networks of Depends and Imports are trees. This makes sense since there may not be cyclical dependencies or imports. The set of nodes of both networks is a subset of the installed packages.
• The network of Suggests has cycles. The set of nodes is larger than the one of installed packages, indicating that some suggested packages are not installed.

Plotting the networks

I am defining a plot_network_packages function to plot the networks. Note how I am choosing a very low value of transparency alpha for edges, as networks are relatively dense.

plot_network_packages <- function(i){
  ggraph(network_packages[[i]], layout = "sugiyama") +
  geom_node_point(aes(label = name)) +
  geom_edge_link(alpha = 0.1, start_cap = circle(3, 'mm'), end_cap = circle(3, 'mm'), arrow = arrow(length = unit(2, 'mm'))) +
  theme_graph() +
  labs(title = paste("Network of", names(network_packages)[i]))
}

network_plots <- lapply(1:3, plot_network_packages)

To plot all networks at once, I am using wrap_plots from the patchwork package. Functions from gridExtra do not seem to work well with ggraph outcomes.

wrap_plots(network_plots, ncol = 1)

We observe that Depends and Imports have a tree like structure, more complex for Imports. Although the Suggests network has cycles and disconnected components, seems to behave like a tree for most of its nodes.

Relevant nodes

Seeing that the networks of packages have a tree-like structure, we can establish two criteria to select relevant nodes:

• Root nodes: the packages at the top of the tree seem to be critical for the functionality of the system. Root nodes will have in-degree equal to zero, and high values of out degree.
• Intermediate nodes: for paths of relationships of two or more edges, the nodes in the middle are also important for package functionality. These nodes will have high values of betweenness.

Let’s create a function to present the relevant packages of a network.

get_table_measure <- function(i, nnodes = 5){
  
  node_table <- network_packages[[i]] %>%
    activate(nodes) %>%
    as_tibble()
  
  root <- node_table %>%
    filter(d_in == 0) %>%
    arrange(-d_out) %>%
    mutate(relation = names(network_packages)[i]) %>%
    select(relation, name, d_out) %>%
    rename(root = name) %>%
    slice(1:nnodes)
  
  interm <- node_table %>%
    arrange(-btw) %>%
    select(name, btw) %>%
    rename(intermediate = name) %>%
    slice(1:nnodes)
  
  table <- bind_cols(root, interm)
  
  return(table)
  
}

Here is the result of applying the function, presented in a table formatted with kableExtra.

nodes_list <- lapply(1:3, \(i) get_table_measure(i))
nodes_table <- bind_rows(nodes_list)

nodes_table %>%
  kbl() %>%
  kable_paper(full_width = FALSE) %>%
  row_spec(1:5, background = "#FFFFCC") %>%
   row_spec(6:10, background = "#CCFFFF") %>%
   row_spec(11:15, background = "#FFCCCC")

Root and intermediate packages are different for each network, representative of each relationship. The main root package in Depends is the minimal version of R required. This was to be expected, given the role of Depends on package dependencies definition. Root packages in Imports are related with the tidyverse like magrittr or with data visualization like grDevices. Root packages in Suggests are related with package development, and most of them are not installed in my computer.

As for intermediate packages, the ones in Depends are not quite representative, as this relationship is not intended to be chained at several levels. The results of Imports are more informative, and show the relevance of the tidyverse in package development, at least the ones in my computer. This is also evident in Suggests, where together with package development appear other packages related with publishing and visualization like knitr and ggplot2.

The results of this analysis are not representative of the whole R CRAN package ecosystem. They have to be considered as a preliminary analysis for a further examination of the whole CRAN network.

References

• Depends or imports? (From the Developing R packages course) https://campus.datacamp.com/courses/developing-r-packages/checking-and-building-r-packages?ex=8
• Dependencies: What does your package need? (from Wickham, H. & Bryant, J. R packages) https://r-pkgs.org/description.html#description-dependencies

Session info

## R version 4.2.0 (2022-04-22)
## Platform: x86_64-pc-linux-gnu (64-bit)
## Running under: Linux Mint 19.2
## 
## Matrix products: default
## BLAS:   /usr/lib/x86_64-linux-gnu/openblas/libblas.so.3
## LAPACK: /usr/lib/x86_64-linux-gnu/libopenblasp-r0.2.20.so
## 
## locale:
##  [1] LC_CTYPE=es_ES.UTF-8       LC_NUMERIC=C              
##  [3] LC_TIME=es_ES.UTF-8        LC_COLLATE=es_ES.UTF-8    
##  [5] LC_MONETARY=es_ES.UTF-8    LC_MESSAGES=es_ES.UTF-8   
##  [7] LC_PAPER=es_ES.UTF-8       LC_NAME=C                 
##  [9] LC_ADDRESS=C               LC_TELEPHONE=C            
## [11] LC_MEASUREMENT=es_ES.UTF-8 LC_IDENTIFICATION=C       
## 
## attached base packages:
## [1] stats     graphics  grDevices utils     datasets  methods   base     
## 
## other attached packages:
## [1] kableExtra_1.3.4  patchwork_1.1.1   ggraph_2.0.5      ggplot2_3.3.5    
## [5] tidygraph_1.2.1   dplyr_1.0.9       data.table_1.14.2
## 
## loaded via a namespace (and not attached):
##  [1] tidyselect_1.1.2   xfun_0.30          bslib_0.3.1        purrr_0.3.4       
##  [5] graphlayouts_0.8.0 colorspace_2.0-3   vctrs_0.4.1        generics_0.1.2    
##  [9] viridisLite_0.4.0  htmltools_0.5.2    yaml_2.3.5         utf8_1.2.2        
## [13] rlang_1.0.2        jquerylib_0.1.4    pillar_1.7.0       glue_1.6.2        
## [17] withr_2.5.0        DBI_1.1.2          tweenr_1.0.2       lifecycle_1.0.1   
## [21] stringr_1.4.0      munsell_0.5.0      blogdown_1.9       gtable_0.3.0      
## [25] rvest_1.0.2        evaluate_0.15      labeling_0.4.2     knitr_1.39        
## [29] fastmap_1.1.0      fansi_1.0.3        highr_0.9          Rcpp_1.0.8.3      
## [33] scales_1.2.0       webshot_0.5.3      jsonlite_1.8.0     systemfonts_1.0.4 
## [37] farver_2.1.0       gridExtra_2.3      ggforce_0.3.3      digest_0.6.29     
## [41] stringi_1.7.6      bookdown_0.26      ggrepel_0.9.1      polyclip_1.10-0   
## [45] grid_4.2.0         cli_3.3.0          tools_4.2.0        magrittr_2.0.3    
## [49] sass_0.4.1         tibble_3.1.6       crayon_1.5.1       tidyr_1.2.0       
## [53] pkgconfig_2.0.3    ellipsis_0.3.2     MASS_7.3-57        xml2_1.3.3        
## [57] svglite_2.1.0      httr_1.4.2         viridis_0.6.2      assertthat_0.2.1  
## [61] rmarkdown_2.14     rstudioapi_0.13    R6_2.5.1           igraph_1.3.1      
## [65] compiler_4.2.0