Non-core rate by creation-year cohort and fixed repo age

For each source, bin non-core issue activity by (creation-year cohort, age-in-years-since-creation) instead of by (popularity stratum, calendar month). This is the age/period/cohort disentangling trick: fixing "age" (e.g. a repo's first or second year of life) while letting "cohort" (the calendar year that age happened to fall in) vary isolates a shared calendar-time effect from a repository's own maturation curve. Pure maturation predicts roughly constant rates across cohorts at a fixed age; a shared calendar-time break instead predicts a monotonic decline in that fixed-age rate as cohort year increases.

Usage

cohort_age_rate_tbl(
  issue_authors_tbl,
  repo_tbl,
  sources,
  contrib_threshold = 0.01,
  metric = c("issues", "comments"),
  age_years = 0:1,
  date_start = as.Date("2010-01-01")
)

Arguments

issue_authors_tbl

As returned by fetch_issue_authors().

repo_tbl

As returned by build_repo_tbl().

sources

Character vector of source_name values to include.

contrib_threshold, metric

As in issue_rate_tbl().

age_years

Integer vector of ages (in whole years since repo creation) to compute rates for. Default 0:1 (a repo's first and second year of life).

date_start

Earliest calendar month considered when accumulating repo-month exposure (not an age cutoff).

Value

A tibble: source, cohort_year, age_year, n_metric, n_repo_months, rate.

Examples

if (FALSE) { # \dontrun{
cohort_tbl <- cohort_age_rate_tbl (issue_authors_tbl, repo_tbl, "cran")
plot_cohort_age (cohort_tbl)
} # }