center_of_mass.R

library(tidyverse)
df <- read_csv("https://raw.githubusercontent.com/LingConLab/rutul_dialectology/master/data/database.csv")
df |> 
  select(feature_title, feature_lexeme, value, settlement, value) |> 
  filter(!is.na(value),
         value != "NO DATA",
         value != "OTHER",
         value != "\\?",
         !(settlement %in% c("Tsudik", "Borch"))) |> 
  mutate(value = str_split(value, ";")) |> 
  unnest_longer(value) |> 
  mutate(value = str_squish(value)) |> 
  group_by(feature_title, feature_lexeme, settlement) |> 
  ungroup() |> 
  arrange(feature_title, feature_lexeme, settlement)  |> 
  distinct(settlement, value, feature_title, feature_lexeme) |> 
  mutate(feature_lexeme = ifelse(is.na(feature_lexeme), "", feature_lexeme),
         merged_value = str_c(feature_title, feature_lexeme, value)) |> 
  arrange(settlement) |> 
  select(settlement, merged_value) |> 
  group_by(merged_value) |> 
  mutate(value = 1,
         shared_with_n_villages = sum(value)) |> 
  ungroup() |> 
  count(settlement, shared_with_n_villages) |> 
  filter(shared_with_n_villages < 12) |> 
  group_by(settlement) |> 
  mutate(cumsum = cumsum(n),
         median = max(cumsum)/2,
         more_then_median = if_else(cumsum >= median, "less or equal", "more")) |> 
  ggplot(aes(shared_with_n_villages, settlement, label = n))+
  geom_tile(aes(fill = n), colour = "white") +
  geom_text(aes(label = n, color = more_then_median), show.legend = FALSE) +
  scale_fill_gradient(low = "lightblue", high = "navy")+
  coord_fixed()+
  labs(x = NULL, y = NULL)+
  scale_x_continuous(breaks = 1:11)+
  theme_minimal()