In openpharma/GithubMetrics: Quickly Get Key Metrics on Github Repositories

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)

library(dplyr)
library(GithubMetrics)

Intro

The following vignette shows how the OS Health is calculated, with the aim of capturing the current vibrancy of the packages community, and their activity on the repo.

It is important to note that 'OS Health' is not equivalent to metrics assessed in riskmetric. An example being a stable and mature statistical R package that could be orphaned, or stable enough activity has largely ended. In terms of 'OS Health', the developer community is not active - but the package itself could still be robust and meeting all it's intended use cases (OS Health and the scope of riskmetric would overlap though if the package has stale open issues that are being ignored).

Note - in the context of openpharma.github.io and the pharmaverse.org, we may deprecate using this package and move to GrimoreLab tooling.

Get data

# repos
  repos <- c(
    "Roche/crmPack" ,
    "Roche/rtables",
    "openpharma/visR",
    "openpharma/GithubMetrics",
    "pharmaverse/admiral"
  )

# commits
  commits <- gh_commits_get(
    repos,
    days_back = 365*10
  )

# issues
  issues <- gh_issues_get(repos, days_back =  365*10)

Create metrics

issues_enriched <- issues %>%
  mutate(
    days_open = as.numeric(Sys.Date() - as.Date(created)),
    days_no_activity = as.numeric(Sys.Date() - as.Date(updated))
  ) %>%
  select(
    full_name, state, days_open, days_no_activity
  )

commits_enriched <- commits %>%
  mutate(
    date = as.Date(datetime)
  ) %>%
  select(full_name, date, author)

scoring <- tibble(
  full_name = unique(commits$full_name)
  ) %>%
  left_join(
    gh_metric_issues(issues_enriched), by = "full_name"
  ) %>%
  left_join(
    gh_metric_commits_days_since_commit(commits_enriched), by = "full_name"
  ) %>%
  left_join(
    gh_metric_commits_prepost_midpoint(commits_enriched), by = "full_name"
  ) %>%
  left_join(
    gh_metric_commits_authors_ever(commits_enriched), by = "full_name"
  ) %>%
  left_join(
    gh_metric_commits_authors_prepost_midpoint(commits_enriched), by = "full_name"
  ) %>%
  gh_score()

Results

These results were generated at r Sys.time().

scoring %>% select(full_name, score) %>%
  knitr::kable(
    caption = "OS Health Score"
  )

Negative factors lowering OS health.

for (i in 1:nrow(scoring)){
  cat(glue::glue("

*{scoring[i,]$full_name}*

* Score is {scoring[i,]$score}  
{scoring[i,]$warnings}


  "))
}

openpharma/GithubMetrics documentation built on Dec. 1, 2023, 12:52 a.m.