Tidy Tuesday Week 28/2021

Tidy Tuesday week 28 Independence Days, data from Wikipedia.

# load libraries
library(tidyverse)
library(ggtext)
library(lubridate)
library(colorspace)
library(maps)
library(patchwork)

# theme set
theme_set(theme_minimal(base_size = 10))

# import data
holidays <- readr::read_csv('https://raw.githubusercontent.com/rfordatascience/tidytuesday/master/data/2021/2021-07-06/holidays.csv')

── Column specification ─────────────────────────────────────────────────────────────────────────
cols(
  country = col_character(),
  date_parsed = col_date(format = ""),
  weekday = col_character(),
  day = col_double(),
  month = col_character(),
  name_of_holiday = col_character(),
  date_of_holiday = col_character(),
  year_of_event = col_double(),
  independence_from = col_character(),
  event_commemorated_and_notes = col_character(),
  year = col_double(),
  date_mdy = col_character()
)

glimpse(holidays)

Rows: 216
Columns: 12
$ country                      <chr> "Afghanistan", "Albania", "Algeria", "Andorra", "Angola", …
$ date_parsed                  <date> 1919-08-19, 1912-11-28, 1962-07-05, NA, 1975-11-11, 1981-…
$ weekday                      <chr> "Tuesday", "Thursday", "Thursday", NA, "Tuesday", "Sunday"…
$ day                          <dbl> 19, 28, 5, NA, 11, 1, 9, 28, 21, NA, 26, 28, 18, 10, 16, 2…
$ month                        <chr> "Aug", "Nov", "Jul", NA, "Nov", "Nov", "Jul", "May", "Sep"…
$ name_of_holiday              <chr> "Afghan Independence Day (Afghan Victory Day)", "Flag Day …
$ date_of_holiday              <chr> "August 19", "November 28", "July 5", NA, "November 11", "…
$ year_of_event                <dbl> 1919, 1912, 1962, NA, 1975, 1981, NA, NA, 1991, NA, 1955, …
$ independence_from            <chr> "United Kingdom", "Ottoman Empire", "France", NA, "Portuga…
$ event_commemorated_and_notes <chr> "Anglo-Afghan Treaty of 1919 or Treaty of Rawalpindi, an a…
$ year                         <dbl> 1919, 1912, 1962, NA, 1975, 1981, 1816, 1918, 1991, NA, 19…
$ date_mdy                     <chr> "August 19, 1919", "November 28, 1912", "July 5, 1962", NA…

# number of countries in dataset
n_distinct(holidays$country)

[1] 195

Independence event count: year

• shared on Twitter

# independence year
ind_year = holidays %>% count(year) %>%
  drop_na() 

lab = ind_year %>% arrange(desc(n)) %>% slice(1:4)

ind_year %>%
  ggplot(aes(x=year, y=n)) + 
  geom_segment(aes(x=year, xend=year, y=0, yend=n, color=I(ifelse(n==max(n),"#9d0208","black")))) +
  geom_point(aes(size=n, color=I(ifelse(n==max(n),"#9d0208","black")))) + 
  geom_text(data = lab,aes(label=n), color="white",size=3) +
  annotate(geom="text",x=1291, y=2.2, label="Switzerland",hjust = "left",size=2.2, angle=50) + 
  annotate(geom="text",x=1523, y=2.2, label="Sweden",hjust = "left",size=2.2, angle=50) + 
  annotate(geom="text",x=1581, y=2.2, label="The Netherlands",hjust = "left",size=2.2, angle=50) + 
  annotate(geom="text",x=1635, y=2.2, label="Portugal",hjust = "left",size=2.2, angle=50) +
  annotate(geom="text",x=1655, y=2.2, label="Oman",hjust = "left",size=2.2, angle=50) +
  annotate(geom="text", x= 1780, y=17, 
           label="17 countries in 1960",size=2.7, hjust="left") +
  annotate(
    geom = "curve", x = 1900, y = 17, xend = 1940, yend = 17, 
    curvature = 0, arrow = arrow(length = unit(1.5, "mm"))
  ) +
  theme(panel.grid.minor = element_blank(),
        legend.position = "none",
        axis.title=element_blank(),
        panel.grid.major=element_line(size=0.3),
        plot.margin=ggplot2::margin(0.5,0.5,0.5,0.5,"cm"),
        plot.caption = element_text(size=7.5, color="#495057"),
        plot.title=element_markdown(
          margin=margin(t=15,b=-60),hjust=0,face="bold",size=26,color="black")) + 
  scale_y_continuous(position="right") + 
  scale_x_continuous(breaks=c(1291,1523, 1581, 1640, 1776,1821,1918,2011)) +
  labs(caption= "\nTidy Tuesdy Week 28 | Data from Wikipedia") + 
  ggtitle("195 COUNTRIES<br><span style = 'color:#9d0208'>INDEPENDENCE YEAR</span>")

Countries with 2 independence events

# countries with 2 independence events
c2 = holidays %>% count(country) %>% arrange(desc(n)) %>% filter(n==2)
n_distinct(c2$country)

[1] 21

# map data
world= map_data("world")

# check for differences
setdiff(c2$country, world$region)

[1] "Congo, Republic of the"

# recode 
c2 = c2 %>% mutate(region= ifelse(country=="Congo, Republic of the","Republic of Congo",country))

# join
joined = left_join(world, c2, by="region") %>% 
  mutate(col = ifelse(is.na(n),"0","2"))

# plot
joined %>% ggplot(aes(x = long, y = lat, group = group)) +
  geom_polygon(aes(fill=col)) + 
  coord_fixed(1.3) + 
  theme(legend.position = "none") + 
  scale_fill_manual(values=c("grey90","#ee9b00")) + 
  theme_void() +
  theme(legend.position = "none",
        plot.title = element_text(face="bold",color="#ee9b00",size=12,hjust=0.5),
        plot.margin=ggplot2::margin(0.5,0,0.5,0,"cm")) + 
  labs(title = "21 Countries With Two Independence Events\n")

Independence event count: day of month

# day of month
holidays %>% count(day) %>%
  drop_na() %>%
  ggplot(aes(x=day, y=n)) + 
  geom_segment(aes(x=day, xend=day, y=0, yend=n)) + 
  geom_point(aes(size=n, color=n)) + 
  theme(panel.grid.minor = element_blank(),
        legend.position = "none",
        plot.title.position = "plot",
        plot.subtitle = element_text(face="bold"),
        panel.grid.major=element_line(size=0.3),
        axis.title=element_text(face="bold",size=9),
        plot.margin=ggplot2::margin(0.5,0.5,0.5,0.5,"cm")) + 
  scale_x_continuous(limits=c(1,31), breaks=seq(1,31,10)) + 
  labs(subtitle="Independence event count, by day of month\n", x="Day of month",y="Count") + 
  scale_color_continuous_sequential(palette="Viridis", trans="reverse")

Independence event count: month and week day

# prepare data
holidays2 = holidays %>%
  mutate(date = mdy(date_mdy),
         wday = wday(date, week_start = 1, label=T),
         wday_n = wday(date, week_start = 1),
         mth_n = month(date),
         mth = month(date, label=T)) 

# month 
p1 = holidays2 %>% group_by(mth, mth_n) %>% tally() %>% drop_na() %>%
  ggplot(aes(x=mth_n-0.5, y=n)) + 
  geom_step() + 
  geom_segment(aes(x=mth_n-0.5, xend=mth_n+0.5, y=n, yend=n)) + 
  geom_text(aes(x=mth_n, y=n,label=mth), vjust=2, size=3.2) +
  geom_text(aes(x=mth_n, y=n, label=n), vjust=-1.5, size=3.2) +
  scale_x_continuous(breaks=c(1,4,7,10)) + 
  scale_y_continuous(limits=c(5,31)) + 
  theme_void() + 
  theme(plot.margin=ggplot2::margin(0.5,0.5,0.5,0.5,"cm"),
        plot.subtitle = element_text(face="bold",size=9)) + 
  labs(subtitle="Independence event count by month")

# wday 
p2 = holidays2 %>% group_by(wday, wday_n) %>% tally() %>% drop_na() %>%
  ggplot(aes(x=wday_n-0.5, y=n)) + 
  geom_step() + 
  geom_segment(aes(x=wday_n-0.5, xend=wday_n+0.5, y=n, yend=n)) + 
  geom_text(aes(x=wday_n, y=n,label=wday), vjust=2, size=3.2) +
  geom_text(aes(x=wday_n, y=n, label=n), vjust=-1.5, size=3.2) + 
  scale_y_continuous(limits=c(20,35)) + 
  theme_void() + 
  theme(plot.margin=ggplot2::margin(0.5,0.5,0.5,0.5,"cm"),
        plot.subtitle = element_text(face="bold",size=9)) + 
  labs(subtitle="Independence event count by week day")

# combine plot
p1/
  p2