Skip to content

Package directives and collaboration networks in CRAN

Notifications You must be signed in to change notification settings

ikosmidis/cranly

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Travis-CI Build Status Coverage Status CRAN_Status_Badge Licence

cranly

cranly provides core visualizations and summaries for the CRAN package database. It is aimed mainly as an analytics tool for developers to keep track of their CRAN packages and profiles, as well as those of others, which, at least for me, is proving harder and harder with the rapid growth of the CRAN ecosystem.

The package provides methods for cleaning up and organizing the information in the CRAN package database, for building package directives networks (depends, imports, suggests, enhances, linking to) and collaboration networks, and for computing summaries and producing interactive visualizations from the resulting networks. Network visualization is through the visNetwork package. The package also provides functions to coerce the networks to igraph objects for further analyses and modelling.

Installation

Install the development version from github:

# install.packages("devtools")
devtools::install_github("ikosmidis/cranly")

Collaboration and package directives networks in CRAN

Load cranly as

library("cranly")

The first step in the cranly workflow is to try and “clean-up” the package and author names in the data frame that results from a call to tools::CRAN_package_db()

p_db <- tools::CRAN_package_db()
package_db <- clean_CRAN_db(p_db)

The CRAN database we use is from

attr(package_db, "timestamp")
#> [1] "2022-06-17 11:12:49 BST"

Package directives networks

The package directives network can then be built using

package_network <- build_network(package_db)

package_network can then be interrogated using extractor methods (see, ?package_by). For example, my packages can be extracted as follows

my_packages <- package_by(package_network, "Ioannis Kosmidis")
my_packages
#> [1] "betareg"          "brglm"            "brglm2"           "detectseparation"
#> [5] "enrichwith"       "PlackettLuce"     "profileModel"     "trackeR"         
#> [9] "trackeRapp"

and their sub-network of directives can be summarized in an interactive visualization, a snapshot of which is below

plot(package_network, package = my_packages, title = TRUE, legend = TRUE)

We can also compute package summaries and plot “Top-n” lists according to the various summaries

package_summaries <- summary(package_network)
plot(package_summaries, according_to = "n_imported_by", top = 20)

plot(package_summaries, according_to = "page_rank", top = 20)

Collaboration networks

The collaboration network can also be built using a similar call

author_network <- build_network(package_db, perspective = "author")

and the extractor functions work exactly as they did for the package directives network. For example, my collaboration network results can be summarized as an interactive visualization, a snapshot of which is below

plot(author_network, author = "Ioannis Kosmidis")

“Top-n” collaborators according to various summaries can again be computed

author_summaries <- summary(author_network)
plot(author_summaries, according_to = "n_collaborators", top = 20)

plot(author_summaries, according_to = "n_packages", top = 20)

plot(author_summaries, according_to = "page_rank", top = 20)

Well, the usual suspects…

Package dependence trees

Since version 0.2 cranly includes functions for constructing and working with package dependence tree objects. A package’s dependence tree shows what else needs to be installed with the package in an empty package library with the package, and hence it can be used to + remove unnecessary dependencies that “drag” with them all sorts of other packages + identify packages that are heavy for the CRAN mirrors + produced some neat visuals for the package

For example, the dependence tree of the PlackettLuce R package I am co-authoring is

PL_dependence_tree <- build_dependence_tree(package_network, "PlackettLuce")
plot(PL_dependence_tree)

cranly also implements a package dependence index (see ?summary.cranly_dependence_tree for mathematical details). The closer that is to 0 the “lighter” the package is

summary(PL_dependence_tree)
#> $package
#> [1] "PlackettLuce"
#> 
#> $n_generations
#> [1] 3
#> 
#> $parents
#>  [1] "CVXR"        "igraph"      "Matrix"      "matrixStats" "partykit"   
#>  [6] "psychotools" "psychotree"  "qvcalc"      "R6"          "RSpectra"   
#> [11] "sandwich"   
#> 
#> $dependence_index
#> [1] 0.6107209

Check the package vignettes for a more comprehensive tour of the package and for network visualizations on authors with orders of magnitude larger collaboration networks than mine.

Code of Conduct

Please note that this project is released with a Contributor Code of Conduct. By participating in this project you agree to abide by its terms.