human.R

require(googledrive)
require(dplyr)
require(tibble)
require(tidyr)
require(stringr)
require(ggplot2)
require(stopwords)

human <- tbl_df(read.csv("human.csv", encoding = "UTF-8", stringsAsFactors = F))
colnames(human) <- c("time",LETTERS[1:6])

human_count <- human %>% count(A) %>% arrange(desc(n))
ggplot(human_count, aes(x = A, y = n, fill = n)) + geom_bar(stat="identity") + theme(axis.text.x = element_text(angle = 90, hjust = 1, ))

words_B <- human %>% 
  select(A,B) %>% 
  unnest_tokens(word,B) %>% 
  filter(!word %in% stopwords("english"), str_detect(word, "[a-z]"))

words_C <- human %>% 
  select(A,C) %>% 
  unnest_tokens(word,C) %>% 
  filter(!word %in% stopwords("english"), str_detect(word, "[a-z]"))

words_D <- human %>% 
  select(A,D) %>% 
  unnest_tokens(word,D) %>% 
  filter(!word %in% stopwords("english"), str_detect(word, "[a-z]"))

words_E <- human %>% 
  select(A,E) %>% 
  unnest_tokens(word,E) %>% 
  filter(!word %in% stopwords("english"), str_detect(word, "[a-z]"))

words_F <- human %>% 
  select(A,F) %>% 
  unnest_tokens(word,F) %>% 
  filter(!word %in% stopwords("english"), str_detect(word, "[a-z]"))

##
words_B <- inner_join(words_B %>% mutate(word_conc = paste(A, word)),
                      words_B %>% mutate(word_conc = paste(A, word)) %>% count(word_conc, sort = TRUE)
                      ) %>%
  arrange(desc(n)) %>% 
  select(A, word, n)

head_B <- head(words_B, 50)

plot_B <- ggplot(head_B, aes(A, word, fill = n)) +
  geom_tile(show.legend = FALSE, color="white") +
  geom_text(aes(label = paste0(n, "%")), color = "white") + 
  theme_classic() +
  theme(axis.text.x = element_text(angle = 90, hjust = 1, ))

##
words_C <- inner_join(words_C %>% mutate(word_conc = paste(A, word)),
                      words_C %>% mutate(word_conc = paste(A, word)) %>% count(word_conc, sort = TRUE)
) %>%
  arrange(desc(n)) %>% 
  select(A, word, n)

head_C <- head(words_C, 50)

plot_C <- ggplot(head_C, aes(A, word, fill = n)) +
  geom_tile(show.legend = FALSE, color="white") +
  geom_text(aes(label = paste0(n, "%")), color = "white") + 
  theme_classic() +
  theme(axis.text.x = element_text(angle = 90, hjust = 1, ))

##
words_D <- inner_join(words_D %>% mutate(word_conc = paste(A, word)),
                      words_D %>% mutate(word_conc = paste(A, word)) %>% count(word_conc, sort = TRUE)
) %>%
  arrange(desc(n)) %>% 
  select(A, word, n)

head_D <- head(words_D, 50)

plot_D <- ggplot(head_D, aes(A, word, fill = n)) +
  geom_tile(show.legend = FALSE, color="white") +
  geom_text(aes(label = paste0(n, "%")), color = "white") + 
  theme_classic() +
  theme(axis.text.x = element_text(angle = 90, hjust = 1, ))

##
words_E <- inner_join(words_E %>% mutate(word_conc = paste(A, word)),
                      words_E %>% mutate(word_conc = paste(A, word)) %>% count(word_conc, sort = TRUE)
) %>%
  arrange(desc(n)) %>% 
  select(A, word, n)

head_E <- head(words_E, 50)

plot_E <- ggplot(head_E, aes(A, word, fill = n)) +
  geom_tile(show.legend = FALSE, color="white") +
  geom_text(aes(label = paste0(n, "%")), color = "white") + 
  theme_classic() +
  theme(axis.text.x = element_text(angle = 90, hjust = 1, ))

##

words_F <- inner_join(words_F %>% mutate(word_conc = paste(A, word)),
                      words_F %>% mutate(word_conc = paste(A, word)) %>% count(word_conc, sort = TRUE)
) %>%
  arrange(desc(n)) %>% 
  select(A, word, n)

head_F <- head(words_F, 50)

plot_F <- ggplot(head_F, aes(A, word, fill = n)) +
  geom_tile(show.legend = FALSE, color="white") +
  geom_text(aes(label = paste0(n, "%")), color = "white") + 
  theme_classic() +
  theme(axis.text.x = element_text(angle = 90, hjust = 1, ))


##

filename = paste0("plot_B", ".png")
ggsave(
  filename = filename,
  plot = plot_B,
  device = "png",
  width = 25,
  height = 25,
  units = "cm",
  dpi = 500
)


filename = paste0("plot_C", ".png")
ggsave(
  filename = filename,
  plot = plot_C,
  device = "png",
  width = 25,
  height = 25,
  units = "cm",
  dpi = 500
)


filename = paste0("plot_D", ".png")
ggsave(
  filename = filename,
  plot = plot_D,
  device = "png",
  width = 25,
  height = 25,
  units = "cm",
  dpi = 500
)


filename = paste0("plot_E", ".png")
ggsave(
  filename = filename,
  plot = plot_E,
  device = "png",
  width = 25,
  height = 25,
  units = "cm",
  dpi = 500
)


filename <- paste0("plot_F", ".png")
ggsave(
  filename = filename,
  plot = plot_F,
  device = "png",
  width = 25,
  height = 25,
  units = "cm",
  dpi = 500
)