In mpadge/pkgstats-analyses: Analyses of 'pkgstats' Results for all CRAN Packages

library ("pkgstatsAnalyses")
here <- here::here ()
v_data_dir <- file.path (here, "vignettes", "data")
logmean <- function (x, ...) {
    10 ^ mean (log10 (x [which (x > 0)]), na.rm = TRUE)
}

datafile <- file.path (here, "data-raw", "pkgstats-results.Rds")
fig07_png <- file.path (here, "vignettes", "figures", "_fig07.png")
f <- file.path (v_data_dir, "fig07_deps_p1.Rds")
calc_fig7_p1_deps <- !file.exists (f) & !file.exists (fig07_png)

recommended <- recommended_pkgs ()
x <- load_pkgstats_data (datafile, raw = TRUE, latest = FALSE)

f_cpl <- file.path (v_data_dir, "fig07-coupling_deps.Rds")
if (!file.exists (f_cpl)) {
    deps <- coupling_dependencies (x)
    saveRDS (deps, f_cpl)
}
deps <- readRDS (f_cpl)

deps_p1 <- deps |>
    group_by (from, version) |>
    summarise (n_base = n_unique [to == "base"],
               n_recommended = sum (n_unique [to %in% recommended]),
               n_contrib = sum (n_unique [!to %in% c ("base", recommended)]),
               .groups = "keep") |>
    mutate (n_total = n_base + n_recommended + n_contrib,
            n_base = n_base / n_total,
            n_recommended = n_recommended / n_total,
            n_contrib = n_contrib / n_total) |>
    group_by (from) |>
    mutate (seq = seq_along (n_base)) |>
    mutate_at(vars (n_base, n_recommended, n_contrib),
              ~replace (., !is.finite (.), NA)) |>
    group_by (seq) |>
    summarise (n_base = mean (n_base, na.rm = TRUE),
               n_recommended = mean (n_recommended, na.rm = TRUE),
               n_contrib = mean (n_contrib, na.rm = TRUE)) |>
    pivot_longer (cols = c (n_base, n_recommended, n_contrib)) |>
    filter (is.finite (value))

saveRDS (deps_p1, f)

readRDS (f) |>
    filter (seq <= 50) |>
    ggplot (aes (x = seq, y = value, colour = name)) +
        geom_line (lty = 1) +
        #geom_smooth (method = "loess", formula = "y ~ x", se = FALSE, lwd = 0.5) +
        theme (legend.title = element_blank(),
               legend.position = c (0.30, 0.6),
               legend.background = element_rect(fill='transparent', colour='transparent')) +
        xlab ("Release sequence") +
        ylab ("Proportion") +
        ggtitle ("A: Change in calls") -> p1

# couplings_releases data are calculated in the main pkg fn
# `couplings_releases` in R/couplings.R, and uploaded as binary to GH release.
f_cpl_rel <- file.path (here, "data-raw", "couplings_releases.Rds")
if (!file.exists (f_cpl_rel)) {
    u <- paste0 ("https://github.com/mpadge/pkgstats-analyses/",
                 "releases/download/v0.0.1/",
                 "couplings_releases.Rds")
    download.file (u, f_cpl_rel)
}

# unique & total instabilities are virtually identical
deps_p2 <- readRDS (f_cpl_rel)
#deps_p2$efferent_unique [deps_p2$efferent_unique == 0L] <- NA_integer_
deps_p2 <- deps_p2 |>
    group_by (seq) |>
    summarise (instability = logmean (instability_unique, na.rm = TRUE),
               efferent = logmean (efferent_unique, na.rm = TRUE))
coeff_p2 <- max (deps_p2$efferent, na.rm = TRUE) / max (deps_p2$instability, na.rm = TRUE)
colours_p2 <- c ("instability" = "#F8766D",
                 "efferent" = "#00BFC4")
deps_p2 |>
    filter (seq <= 100) |>
    ggplot (aes (x = seq)) +
        geom_line (aes (y = instability, col = "instability")) +
        geom_line (aes (y = efferent / coeff_p2, col = "efferent")) +
        scale_y_continuous (
            name = "instability",
            sec.axis = sec_axis (~.*coeff_p2, name = "Efferent calls")) +
        xlab ("Release sequence") +
        scale_colour_manual (values = colours_p2) +
        theme (legend.title = element_blank(),
           legend.position = c (0.80, 0.8),
           legend.background = element_rect(fill='transparent', colour='transparent')) +
        ggtitle ("B: Coupling instability") -> p2

# afferent fn imports by other pkgs
deps_p3 <- readRDS (f_cpl_rel)
deps_p3$afferent_fns_per_pkg [which (is.na (deps_p3$afferent_fns_per_pkg))] <- 0L
deps_p3 <- deps_p3 |>
    group_by (seq) |>
    summarise (packages = logmean (afferent_npkgs),
               fns_per_pkg = logmean (afferent_fns_per_pkg)) |>
    filter (seq <= 50)
coeff_p3 <- max (deps_p3$fns_per_pkg) / max (deps_p3$packages)
# colours from `ggplot_build(p1)$data`:
colours_p3 <- c ("fns_per_pkg" = "#F8766D",
                 "packages" = "#00BFC4")
p3 <- ggplot (deps_p3, aes (x = seq)) +
        geom_line (aes (y = packages, col = "packages")) +
        geom_line (aes (y = fns_per_pkg / coeff_p3, col = "fns_per_pkg")) +
        scale_y_continuous (
            name = "packages",
            sec.axis = sec_axis (~.*coeff_p3, name = "Imported fns / pkg")) +
        xlab ("Release sequence") +
        scale_colour_manual (values = colours_p3) +
        theme (legend.title = element_blank(),
               legend.position = c (0.15, 0.5),
               legend.background = element_rect(fill='transparent', colour='transparent')) +
        ggtitle ("C: Imports by other pkgs")

png (fig07_png, width = 7, height = 7, units = "in", res = 300)
(p1 | p2) / p3
graphics.off ()

knitr::include_graphics (fig07_png)

cp <- readRDS (f) |>
    group_by (seq) |>
    slice_max (seq)
x <- load_pkgstats_data (datafile, raw = TRUE, latest = FALSE) |>
    group_by (date) |>
    slice_max (date) |>
    ungroup () |>
    select (package, n_fns_r_exported)
cp <- left_join (cp, x, by = "package")
tmp <- data.frame (aff = log10 (cp$afferent_unique),
                   nfns = log10 (cp$n_fns_r_exported)) |>
    filter (is.finite (aff) & is.finite (nfns))
ggplot (tmp, aes (x = nfns, y = aff)) +
    geom_hex () +
    geom_smooth (method = "lm", formulat = "y ~ x")
mod <- lm (aff ~ nfns, data = tmp)