Analysis Stark Trek Commands

TidyTuesday Project (210817)

• 📱 Linkedin Juan Antonio Cabeza Sousa
• 📬 Email: juaancabsou@gmail.com
• 🖥️ Twitter: @Aceconhielo

This notebook is about Analysing Star Trek Commands.The data this week comes from SpeechInteraction.org.

A glance to the data

Looking the Data Dictionary related to this data set and the amount of missing values per column, it helps us to keep just a few important columns for the visualization stage. Those columns (the ones that I will kept for the TidyTuesday visualization) are: * char: Name of the speaking character. * char_type:Either Person or Computer * line: The complete line of dialog (may contain more speech than the speech interaction). Parenthetical are directions and not spoken. * type: The type of interaction, see detailed definitions below. * pri_type: The primary interaction type as defined by the below ranking. * domain: The domain of interaction, see detailed definitions below. * sub_domain: The sub-domain of interaction, generally a specific setting, see below. * interaction: The actual speech interaction. May be shorter or longer than the line of dialog.

I will discard is_fedand error because their distribution is always the same: TRUE and FALSE respectively.

ggarrange(chart_intro, chart_miss, ncol = 2, heights = c(10, 10))

# Top 10 speakers: included computer voice
talkers <- head(
  computer %>% 
  group_by(char) %>%
  count(sort = TRUE),10)$char

# Chart 1: Interactions
interactions <- head(computer %>% filter(char != 'Computer Voice') %>%
  group_by(char) %>%
  count(sort = TRUE),10)

interactions$char <- factor(interactions$char, levels = interactions$char[order(interactions$n)])
colnames(interactions) <- c('Character', 'Interactions')

# Chart 2: Wordclouds
textcleaned <- computer %>%
  unnest_tokens(word, interaction) %>%
  anti_join(get_stopwords(source = "snowball"))

Joining, by = "word"

df_wordcloud <- textcleaned %>% group_by(word) %>% count(sort = TRUE)

# Chart 3: Waffles
interactions_pct <- computer %>%
  filter(char %in% talkers) %>%
  group_by (char, type) %>%
  summarise (n=n()) %>%
  mutate(rel.freq = paste0(round(100 * n/sum(n), 0), "%"))

`summarise()` has grouped output by 'char'. You can override using the `.groups` argument.

backgroundcolor <- "#264653"
solidchartcolor <- "#e9c46a"
gridlines <- "#006d77"


# Chart 1: Interactions
chartinteractions<-
  ggplot(data=interactions, aes(x=Character, y=Interactions)) +
  geom_chicklet(stat="identity", color=solidchartcolor, fill=solidchartcolor, width = 0.75, radius = grid::unit(1, 'mm'))+
  geom_text(aes(label=Interactions), hjust=1.6,vjust= 0.5, color=backgroundcolor, size=3.5)+
  theme(
    axis.text.x = element_text(colour = 'white'),
    axis.title.x = element_text(colour = 'white'),
    axis.text.y = element_text(colour = 'white'),
    axis.title.y = element_text(colour = 'white'),
    panel.background = element_rect(fill = backgroundcolor, colour = backgroundcolor, linetype = "solid"),
    panel.grid.major = element_line(size = 0.5, linetype = 'solid', colour = gridlines), 
    panel.grid.minor = element_line(size = 0.25, linetype = 'solid', colour = gridlines),
    plot.background = element_rect(fill = backgroundcolor),
    plot.title = element_text(colour="white"),
    plot.margin = unit(c(1,1,1,1), "cm")
  )+
  ggtitle("Users with more interactions")+
  coord_flip()

Warning: Ignoring unknown parameters: stat

chartinteractions2<-
  ggplot() +
  theme(
    axis.text.x = element_text(colour = 'white'),
    axis.title.x = element_text(colour = 'white'),
    axis.text.y = element_text(colour = 'white'),
    axis.title.y = element_text(colour = 'white'),
    panel.background = element_rect(fill = backgroundcolor, colour = backgroundcolor, linetype = "solid"),
    panel.grid.major = element_line(size = 0.5, linetype = 'solid', colour = gridlines), 
    panel.grid.minor = element_line(size = 0.25, linetype = 'solid', colour = gridlines),
    plot.background = element_rect(fill = backgroundcolor),
    plot.title = element_text(colour="white"),
    plot.margin = unit(c(1,1,1,1), "cm")
  )+
  coord_flip()

wordcloudchart <- wordcloud(words = df_wordcloud$word ,
                            freq = df_wordcloud$n,
                            min.freq = 1,max.words=200,
                            random.order=FALSE,
                            rot.per=0.35,
                            colors=brewer.pal(8, "Dark2")
                            )

library(waffle)

generatewaffle <- function(df,t,rows){
  wdata <- df %>% filter(char == t)
  wdata
  fig <- ggplot(wdata, aes(fill = type, values = n)) +
  geom_waffle(n_rows = rows, size = 0.33, colour = "white") +
  theme(
    axis.text.x = element_text(colour = 'white'),
    axis.title.x = element_blank(),
    axis.text.y = element_text(colour = 'white'),
    axis.title.y = element_blank(),
    panel.background = element_rect(fill = backgroundcolor, colour = backgroundcolor, linetype = "solid"),
    panel.grid.major = element_line(size = 0.5, linetype = 'solid', colour = backgroundcolor), 
    panel.grid.minor = element_line(size = 0.25, linetype = 'solid', colour = backgroundcolor),
    plot.background = element_rect(fill = backgroundcolor),
    plot.title = element_text(colour="white"),
    legend.background = element_rect(colour = "transparent", fill = backgroundcolor),
    legend.title = element_text(color = backgroundcolor),
    legend.text = element_text(color = "white"),
    legend.key = element_rect(fill = backgroundcolor, color = NA),
  )+
  ggtitle(paste0(t,": Types of interaction"))
  
  return(fig)
}

fig1 <- generatewaffle(interactions_pct, "Computer Voice", 20)
fig2 <- generatewaffle(interactions_pct, "Geordi", 20)
fig3 <- generatewaffle(interactions_pct, "Picard", 20)
fig4 <- generatewaffle(interactions_pct, "Data",10)
fig5 <- generatewaffle(interactions_pct, "Riker", 10)
fig6 <- generatewaffle(interactions_pct, "Beverly", 10)

library(ggpubr)
ggarrange(
  ggarrange(chartinteractions,fig1, ncol=2),
  ggarrange(fig2, fig3, ncol=2),
  ggarrange(fig4, fig5, fig6, ncol=3),
  nrow=3,
  solidchartcolor='red'
)

Warning in as_grob.default(plot) :
  Cannot convert object of class character into a grob.

$`1`

$`2`

attr(,"class")
[1] "list"      "ggarrange"