Voting similarity among Bird of the Year 2019 candidates

botypatterns2019.R

library(Rtsne)
library(vroom)
library(dplyr)
library(tidyr)
library(stringr)
library(ggplot2)
library(ggrepel)
library(janitor)

# data from https://www.dragonfly.co.nz/news/2019-11-12-boty.html
# CC attribution 4 licence

BOTY <- vroom("BOTY-votes-2019.csv", delim = ",")

long <- BOTY %>% mutate(voter=paste0("vote",row_number())) %>%
  gather(vorder, vote, vote_1:vote_5) %>% 
  arrange(voter,vorder)


birds <- data.frame(vote = unique(long$vote),
                    safe = make_clean_names(unique(long$vote)),
                    stringsAsFactors = FALSE) %>%
  filter(!is.na(vote))
longclean <- long %>% inner_join(birds, by = "vote") 

paired_clean <- longclean %>%
  arrange(voter,safe) %>%
  group_by(voter) %>% 
  summarise(x12 = paste(safe[1],safe[2]),
            x13 = paste(safe[1],safe[3]),
            x14 = paste(safe[1],safe[4]),
            x15 = paste(safe[1],safe[5]),
            x23 = paste(safe[2],safe[3]),
            x24 = paste(safe[2],safe[4]),
            x25 = paste(safe[2],safe[5]),
            x34 = paste(safe[3],safe[4]),
            x35 = paste(safe[3],safe[5]),
            x45 = paste(safe[4],safe[5])) %>%
  ungroup() %>%
  gather(vorder, vpair, x12:x45) %>%
  count(vpair, sort=TRUE) %>%
  filter(!str_detect(vpair, fixed("NA")))

results <- data.frame(species = unique(longclean$safe),
                      stringsAsFactors = FALSE)  
for (abird in results$species) {
  addenda <- paired_clean %>% 
    filter(str_detect(vpair, fixed(abird))) %>%
    mutate(species = trimws(gsub(abird, "", vpair))) %>%
    mutate(prop = n/sum(n)) %>% 
    select(species, prop)
  results <- results %>% left_join(addenda, by = "species") %>%
    mutate(prop = ifelse(is.na(prop), 0, prop))
  names(results)[ncol(results)] <- abird
}

tsne <- Rtsne(results[,2:ncol(results)], dims = 2, perplexity=28, verbose=FALSE, theta=0.0, max_iter = 500, check_duplicates=FALSE)
tsne2D <- as.data.frame(tsne$Y)
tsne2D$safe = results$species
df <- tsne2D %>% inner_join(birds, by="safe")

ggplot(df, aes(x=V1,y=V2, label=vote)) + 
  theme_void() +
  ggtitle("2 dimesional (TNSE) representation of similarity by Bird of the Year 2019 covotes") + 
  geom_point(size=1.2) +
  geom_text_repel( size=2.5)