RStudio_ct_selection_untitled_6FA05261

arr <- read_csv("D://arrivals.csv")

arr <- arr %>% 
  mutate(name = replace_na(`Basic data and indicators`, "REMOVE")) %>% 
  select(-`Basic data and indicators`)

tot <- arr %>% 
  filter(X3 == "Total arrivals" | name != "REMOVE") %>% 
  relocate(name)

tot <- tot %>% 
  select(-starts_with("X")) %>% 
  select(-starts_with("X"))

tot$name[tot$name == "REMOVE"] <- NA

tot$name <- zoo::na.locf(tot$name, fromLast=FALSE)

tot <- tot %>% 
  pivot_longer(!name, names_to = "type") 

tot <- tot %>% 
  filter(value != "NA")

tot <- tot %>% 
  mutate(type = as.numeric(type))

tot$name <- quanteda::char_tolower(tot$name)


tot$name <- stringr::str_to_title(tot$name)

tot <- tot %>% 
  rename(year = type) %>% 
  rename(country = name)

cull <- tot %>% 
  group_by(country) %>% 
  summarise(sum = sum(value)) %>% 
  filter(sum >= 500000)

tot <- left_join(tot, cull) %>% 
  filter(sum != "NA")

arr1 <- tot %>% 
  mutate(color = frcode(country == "France" ~ "a",
                        country == "United States Of America" ~ "b",
                        country == "China" ~ "c", 
                        country == "Spain" ~ "d", 
                        country == "Mexico" ~ "e",
                        TRUE ~ "f"))


arr1 %>% 
  ggplot(., aes(x = year, y = value, group = country, color = color)) +
  geom_point() + 
  geom_line() +
  theme_rb(legend = FALSE) +
  add_text(x = 2000, y = 180000, word = "France", sz = 4) +
  add_text(x = 2005, y = 160000, word = "USA", sz = 4) +
  add_text(x = 2015, y = 145000, word = "China", sz = 4) +
  add_text(x = 2012, y = 110000, word = "Spain", sz = 4) +
  add_text(x = 2000, y = 115000, word = "Mexico", sz = 4) +
  scale_color_manual(values = c("#4E79A7", "#F28E2B", "#E15759", "#76B7B2", "#59A14F", "azure3")) +
  labs(x = "", y = "", title = "Number of Arrivals Per Year by Country", caption = "@ryanburge\nData: UNWTO") +
  ggsave("E://tableau_test.png", type = "cairo-png")

# https://git.io/JYm7R