CHI Bot Data Visualization

questions_df$question_text[1]

[1] "Overall, how helpful was the chatbot in helping you articulate events that happened on social media?"

# Frequency of responses
table(pilotdata_1_$helpfulness_chatbot_articulate)

Very unhelpful    Not helpful        Neutral  Quite helpful   Very helpful 
             1              2              7             41             49 

questions_df$question_text[2]

[1] "Overall, how well did the chatbot help you formulate your thoughts around specific social media challenges?"

# Frequency of responses
table(pilotdata_1_$helpfulness_chatbot_formulate)

Not well at all        Not well         Neutral      Quite well       Very well 
              1               1               8              49              41 

# Display the question text
questions_df$question_text[3]

[1] "If you had to take part in this type of study again, would you prefer the story creation process to be:"

# Frequency of responses
table(pilotdata_1_$preferred_story_creation_process)

   survey interview   chatbot 
       34         8        57 

# Load required libraries
library(likert)
library(tidyverse)

# Define recoding mappings for the variables
formulate_levels <- c("Not well at all", "Not well", "Neutral", "Quite well", "Very well")
helpful_levels <- c("Very unhelpful", "Not helpful", "Neutral", "Quite helpful", "Very helpful")


# Create a new structured data frame for helpfulness_chatbot_articulate
helpfulness_data <- data.frame(
  Item = c("Helpfulness of Chatbot"), # The question/item
  `Very unhelpful` = sum(pilotdata_1_$helpfulness_chatbot_articulate == "Very unhelpful", na.rm = TRUE),
  `Not helpful` = sum(pilotdata_1_$helpfulness_chatbot_articulate == "Not helpful", na.rm = TRUE),
  Neutral = sum(pilotdata_1_$helpfulness_chatbot_articulate == "Neutral", na.rm = TRUE),
  `Quite helpful` = sum(pilotdata_1_$helpfulness_chatbot_articulate == "Quite helpful", na.rm = TRUE),
  `Very helpful` = sum(pilotdata_1_$helpfulness_chatbot_articulate == "Very helpful", na.rm = TRUE)
)

# Rename columns for the Likert package
colnames(helpfulness_data) <- c("Item", "Very unhelpful", "Not helpful", "Neutral", "Quite helpful", "Very helpful")

# Use the likert() function with structured data
helpfulness_likert <- likert(summary = helpfulness_data)

# Plot the Likert chart for helpfulness
plot(helpfulness_likert,
     plot.percent.neutral = FALSE, # Suppress neutral percentages
     legend.position = "bottom" # Adjust legend position
) +
theme(aspect.ratio = 0.2) # Adjust the height-to-width ratio

ggsave("helpfulness_plot.png", plot = last_plot(), width = 8, height = 2.1, units = "in", dpi = 300)

# Create a new structured data frame for helpfulness_chatbot_formulate
formulate_data <- data.frame(
  Item = c("Formulate Challenges"), # The question/item
  `Not well at all` = sum(pilotdata_1_$helpfulness_chatbot_formulate == "Not well at all", na.rm = TRUE),
  `Not well` = sum(pilotdata_1_$helpfulness_chatbot_formulate == "Not well", na.rm = TRUE),
  Neutral = sum(pilotdata_1_$helpfulness_chatbot_formulate == "Neutral", na.rm = TRUE),
  `Quite well` = sum(pilotdata_1_$helpfulness_chatbot_formulate == "Quite well", na.rm = TRUE),
  `Very well` = sum(pilotdata_1_$helpfulness_chatbot_formulate == "Very well", na.rm = TRUE)
)

# Rename columns for the Likert package
colnames(formulate_data) <- c("Item", "Not well at all", "Not well", "Neutral", "Quite well", "Very well")

# Use the likert() function with structured data
formulate_likert <- likert(summary = formulate_data)

# Plot the Likert chart for formulate
plot(formulate_likert,
     plot.percent.neutral = FALSE, # Suppress neutral percentages
     legend.position = "bottom" # Adjust legend position
) +
theme(aspect.ratio = 0.2) # Adjust the height-to-width ratio

ggsave("formulate_plot.png", plot = last_plot(), width = 8, height = 2.1, units = "in", dpi = 300)

library(dplyr)
library(ggplot2)
library(RColorBrewer)
library(stringr)

preferred_data <- pilotdata_1_ %>%
  count(preferred_story_creation_process) %>%
  mutate(percentage = n / sum(n) * 100)

# Define a color palette similar to the likert package
likert_colors <- brewer.pal(3, "Set3")

# Plot the pie chart
ggplot(preferred_data, aes(x = "", y = percentage, fill = preferred_story_creation_process)) +
  geom_bar(stat = "identity", width = 1) +
  coord_polar(theta = "y") +
  theme_void() +
  scale_fill_manual(values = likert_colors) +
  labs(title = str_wrap("Preferred Story Creation Process", width = 50), fill = "Process") +
  theme(legend.position = "right")

ggsave("preferred_story_creation_pie.png", width = 6, height = 4, units = "in", dpi = 300)

library(ggplot2)
library(stringr)
# Create Likert objects for visualization
likert_data <- pilotdata_1_ %>% 
  select(helpfulness_chatbot_articulate, helpfulness_chatbot_formulate) %>% 
  rename(
    articulate = helpfulness_chatbot_articulate,
    formulate = helpfulness_chatbot_formulate
  )

# Convert to data frame for likert package
likert_df <- as.data.frame(likert_data)

# Create Likert object
result_likert <- likert(likert_df)

# Plot Likert bar graph
plot(result_likert, type = "bar", centered = FALSE) + 
  ggtitle("Chatbot Helpfulness Ratings")

# Save the plot
ggsave("likert_chatbot_helpfulness.png", width = 8, height = 2.1)