animated-Carpentries-workshops-map.R

library(tidyverse)
library(maps)
library(gganimate)
library(countrycode)

wkshp_data <- read_csv("https://data.softwarecarpentry.org/api/queries/118/results.csv?api_key=GR8itg5dtYtq1A1Myv8bR5rrDNMZRlGc4cMVSnfB") %>%
    filter(start_date < "2019-07-01")


wkshp_data <- wkshp_data %>%
    filter(latitude < 90) %>% 
    mutate(year = lubridate::year(start_date))


wkshp_data_no_online <- wkshp_data %>%
    filter(country != "W3" & 
           !grepl("online", tag_name)) %>%
    mutate(country_name = countrycode(
               country,
               "iso2c",
               "country.name"
           ))

summ_by_country <- wkshp_data_no_online %>%
    count(year, country_name) %>%
    complete(year, country_name, fill = list(n = 0)) %>% 
    group_by(country_name) %>%
    mutate(cum_n = cumsum(n),
           cum_n = replace(cum_n, cum_n == 0, NA))

## first test, only points
ggplot() +
    geom_map(data=world, map=world,
             aes(x = long, y = lat, group = group, map_id = region),
                  fill = "lightgrey", size = 0.5) +
    geom_point(data = wkshp_data, 
               aes(
                   x = longitude,
                   y = latitude,
                   color = as.factor(year))
               ) +
    coord_quickmap() +
    theme_minimal() +
    transition_time(start_date) +
    ease_aes('linear')

## to shade by country, we need to standardize the country names
world <- map_data("world") %>%
    mutate(region_2 = countrycode::countrycode(
                                     region,
                                     "country.name",
                                     "country.name"
                                   ))

res <- map_df(
    unique(summ_by_country$year), 
    function(.x) {
        left_join(world, filter(summ_by_country, year == .x),
                  by = c("region_2" = "country_name")) %>%
            mutate(
                year = replace_na(year, min(year, na.rm = TRUE)),
                start_date = parse_date(
paste0(year, "-12-31"))
            )
    }
)

## and add the columns we need
res <- bind_rows(
    world %>%
    mutate(
        n = NA,
        cum_n = NA, 
        start_date = as.Date("2011-12-31")
    ),
    res
)

anim <- ggplot() +
  geom_map(aes(fill = cum_n, x = long, y = lat, map_id = region),
    data = res, map = world)  +
  scale_fill_viridis_c(na.value = "gray70", 
    breaks = c(1, 10, 100, 500),
    trans = "log", name = "Number of Workshops") +
  geom_point(data = wkshp_data_no_online,
    aes(x = longitude, y = latitude,
                 color = "#dda631"),
    inherit.aes = FALSE) +
  coord_quickmap() +
  theme_minimal() +
  theme(legend.position = "bottom") +
  labs(
    x = "", y = ""
  ) +
  scale_color_identity(guide = FALSE) +
  transition_time(start_date)

## gif version
anim_map <- animate(anim, width = 800, height = 600, res = 144)
anim_save("map-workshop-through-time-animation.gif", anim_map)

### Workshops through time -----------------------------------------------------

wkshp_data <- wkshp_data %>%
    arrange(start_date) %>% 
    mutate(n_wkshp = row_number())

ggplot(wkshp_data, aes(start_date, n_wkshp)) +
  geom_line() +
  geom_point(size = 2) + 
  transition_reveal(start_date) + 
  labs(title = 'Number of Carpentries workshops (since 2012)',
    y = '', x = '') + 
  theme_minimal()

### Waffle plots ---------------------------------------------------------------


wfle <- wkshp_data %>%
  mutate(workshop_type = case_when(
    grepl("SWC", tag_name) ~ "Software Carpentry",
    grepl("DC", tag_name) ~ "Data Carpentry",
    grepl("LC", tag_name) ~ "Library Carpentry",
    grepl("TTT", tag_name) ~ "Instructor Training",
    TRUE ~ "issue"
  )) %>%
  count(year, workshop_type) %>%
  ungroup() %>%
  group_by(year) %>% 
  mutate(tot = sum(n)) %>%
  ungroup() %>%
  mutate(pad = max(tot) - tot)

stopifnot(sum(wfle$workshop_type == "issue") == 0)


## install from GitHub:
## remotes::install_github("hrbrmstr/waffle")
library(waffle)
library(emojifont)

plt <- wfle %>%
  split(.$year) %>%
  purrr::map(function(.x) {
    title <- .x$year[1]
    pad <- .x$pad[1]
    message(str(title))
    cols <- c(
      "Data Carpentry" = "darkcyan",
      "Software Carpentry" = "#2e3c91",
      "Instructor Training" = "#f6a570")[unique(.x$workshop_type)]
    .x <- .x %>%
      select(workshop_type, n) %>%
      arrange(factor(workshop_type,
        c("Software Carpentry", "Data Carpentry", "Instructor Training"))) 
    dt <- pull(.x, n) %>%
      set_names(.x$workshop_type)   
    waffle(dt, rows = 5, pad = pad/5, keep = FALSE,
      size = .5, legend = "none", 
      title = title, colors = cols)
    })

svg(height = 8, width = 11)
iron(plt[[1]], plt[[2]], plt[[3]], plt[[4]],
  plt[[5]], plt[[6]], plt[[7]])
dev.off()

## All workshops 

d <- data.frame(n = c(1, 10, 100), x = 1:3, y = rep(1, 3))

svg("number_workshops.svg", height = 2, width = 6)
wfle %>%
  group_by(year) %>%
  summarize(
    n = sum(n)
  )  %>% 
  ggplot() +
  geom_text(aes(year, I(1), size = n),
    label = fontawesome("fa-users"), 
    family = "fontawesome-webfont") +
  geom_text(aes(year, I(.7), label = n)) +
  geom_text(aes(year, I(.2), label = year)) +
  scale_size(range = c(5, 10)) +
  theme_void() +
  theme(plot.margin = margin(50, 4.5, 4.5, 4.5),  legend.position = "none") +
  coord_cartesian(clip = "off")
dev.off()