Most successful teams of the Copa America!

GanadoresDeCopaAmerica.r

## packages
library(dplyr)
library(tidyr)
library(purrr)
library(stringr)
library(rvest)
library(polite) ## github only
library(ggplot2)
library(scales)
library(glue)
library(extrafont)
loadfonts()

## webscrape
url <- "https://es.wikipedia.org/wiki/Anexo:Estad%C3%ADsticas_de_la_Copa_Am%C3%A9rica"

session <- bow(url)

copa_campeones <- scrape(session) %>% 
  html_nodes(".mw-parser-output > table:nth-child(10)") %>% 
  html_table() %>% 
  flatten_df()

## clean + tidy
copa_campeones_limpia <- copa_campeones %>% 
  janitor::clean_names() %>% 
  slice(1:8) %>%  
  select(1:4) %>% 
  set_names(c("team", "winners", "runners_up", "third_place")) %>% 
  separate(winners, into = c("Champions", "first_place_year"), 
           sep = " ", extra = "merge") %>% 
  separate(runners_up, into = c("Runners-up", "second_place_year"), 
           sep = " ", extra = "merge") %>% 
  separate(third_place, into = c("Third Place", "third_place_year"), 
           sep = " ", extra = "merge") %>% 
  mutate_all(list(~str_replace_all(., "–", "0"))) %>% 
  mutate_at(vars(contains("num")), funs(as.numeric)) %>% 
  gather(key = "key", value = "value", -team, 
         -first_place_year, -second_place_year, -third_place_year) %>% 
  mutate(key = as.factor(key),
         value = as.numeric(value),
         team = team %>% str_replace(., "[A-Z]{3}", "") %>% str_trim(.),
         team = case_when(team == "Brasil" ~ "Brazil",
                          TRUE ~ team)) %>% 
  mutate(key = forcats::fct_relevel(key, 
                                    "Champions",
                                    "Runners-up",
                                    "Third Place")) %>% 
  arrange(key, desc(value)) %>% 
  mutate(team = forcats::as_factor(team),
         order = row_number())

## save data
saveRDS(copa_campeones_limpia, file = here::here("data/copa_campeones_clean.RDS"))
copa_campeones_limpia <- readRDS(file = here::here("data/copa_campeones_clean.RDS"))

## PLOT
copa_ganadores_plot <- copa_campeones_limpia %>% 
  ggplot(aes(value, forcats::fct_rev(team), color = key)) +
  geom_point(size = 10) +        # 10
  geom_text(aes(label = value), 
            size = 5, color = "black",    # 5 
            family = "Roboto Condensed", fontface = "bold") +
  scale_color_manual(values = c("Champions" = "#FFCC33",
                                "Runners-up" = "#999999",
                                "Third Place" = "#CC6600"),
                     guide = FALSE) +
  scale_x_continuous(breaks = c(1, 5, 10, 15),
                     labels = c(1, 5, 10, 15),
                     limits = c(-1, 16)) +
  labs(x = "Number of Occurrence", y = NULL,
       title = "Most Successful Teams of the Copa América!",
       subtitle = str_wrap("Ordered by number of Copa América(s) won. Argentina missed the chance to leapfrog Uruguay after consecutive final losses in the previous two tournaments!", width = 80),
       caption = glue("
                      Source: Wikipedia
                      By @R_by_Ryo")) +
  facet_wrap(~key) +
  theme_copaAmerica(caption.size = 10)

copa_ganadores_plot