analysis3.qmd

---
title: "Analysis for Transfer Experiment"
format: html
editor: visual
---


Packages.

```{r}
library(here)
library(tidyverse)
library(janitor)
library(lme4)
library(ggthemes)
```

```{r}
pilotA_data <- read_csv(here("data","amg_gamelogs_socialtransfer_experiment_pilota.csv")) 
pilotA_data <- pilotA_data |> 
  filter(game_type != 'practice') |>
  filter(game_type != 'complete') |>
  filter(prolific_id != 'julio_test') |>
  group_by(prolific_id, session_index, session_id, game_type,
           game_sequence, teacher, transfer, block_index) |>
  summarise(total_reward = max(game_total_reward)) |>
  ungroup()
  

pilotA_data


# Calculate mean and standard error for total_reward
data_summary <- pilotA_data %>%
  group_by(transfer, teacher, game_sequence, block_index) %>%
  summarise(
    mean_reward = mean(total_reward),
    se_reward = sd(total_reward) / sqrt(n()),  # Standard error
    .groups = 'drop'  # This ensures the result is not grouped
  )

data_summary
```


```{r}
# Create the plot
ggplot(data_summary, aes(x = block_index, y = mean_reward, group = teacher, color = teacher)) +
  geom_line() +  # Line for mean rewards
  geom_errorbar(aes(ymin = mean_reward - se_reward, ymax = mean_reward + se_reward), width = 0.1) +
  facet_grid(game_sequence ~ transfer, scales = "free_x") +  # Organizing plots by game sequence and transfer
  scale_color_manual(values = c("expert" = "blue", "novice" = "red", "none" = "green")) +
  labs(title = "Game Reward Analysis (Mean and SE)",
       x = "Game Index",
       y = "Mean Total Reward") +
  theme_minimal() +
  theme(strip.background = element_blank(), 
        strip.text.x = element_text(size = 12, face = "bold"), 
        strip.text.y = element_text(size = 12, face = "bold"),
        legend.position = "bottom",
        legend.title = element_blank())
```



```{r}

# Mapping block_index values
data_summary$block_index_mapped <- dplyr::recode(data_summary$block_index,
                                                 `0` = 24,
                                                 `1` = 48,
                                                 `2` = 52,
                                                 `3` = 76)

# Create the plot with the adjusted facet grid layout using the mapped block_index

#data_summary$transfer <- factor(data_summary$transfer, labels = c("Transfer: True", "Transfer: False"))

# Correctly assigning factor levels and labels
data_summary$transfer <- factor(data_summary$transfer, levels = c(TRUE, FALSE), labels = c("Transfer: True", "Transfer: False"))


p <- ggplot(data_summary, aes(x = block_index_mapped, y = mean_reward, group = teacher, color = teacher)) +
  geom_line() +  # Line for mean rewards
  geom_errorbar(aes(ymin = mean_reward - se_reward, ymax = mean_reward + se_reward), width = 0.1) +
  facet_grid(transfer ~ game_sequence, scales = "free_x") +  # Adjusted for Transfer on rows and Game Sequence on columns
  scale_color_manual(values = c("expert" = "blue", "novice" = "red", "none" = "green")) +
  scale_x_continuous(breaks = c(24, 48, 52, 76), labels = c("24", "48", "52", "76")) +  # Set custom x-axis ticks
  labs(title = "Game Reward Analysis (Mean and SE)",
       x = "Number of Trials (Shots)",
       y = "Mean Total Reward") +
  theme_minimal() +
  theme(strip.background = element_blank(), 
        strip.text.x = element_text(size = 12, face = "bold"), 
        strip.text.y = element_text(size = 12, face = "bold"),
        legend.position = "bottom",
        legend.title = element_blank())


ggsave(here("figures", "exploratory_plot1.png"), plot = p, width = 7.5, height = 5)

# Display the plot
p
```
```{r}
pilotB_prolific_data <- read_csv(here("data","socialtransferbandits_pilotb_prolific_export.csv"))
pilotB_prolific_data

pilotB_prolific_data_filtered <- pilotB_prolific_data |>
  dplyr::filter(!(Status == "RETURNED" | grepl("^Yes", Colourblindness)))

pilotB_prolific_data_filtered
```

```{r}
pilotB_data <- read_csv(here("data","socialtransferbandits_pilotb_amg_gamelogs.csv")) 
pilotB_data <- pilotB_data |>
  filter(prolific_id %in% pilotB_prolific_data_filtered$`Participant id`) |>
  filter(game_type != 'practice') |>
  filter(game_type != 'complete') |>
  filter(prolific_id != 'julio_test') |>
  group_by(prolific_id, session_index, session_id, game_type,
           game_sequence, teacher, transfer, block_index) |>
  summarise(total_reward = max(game_total_reward)) |>
  ungroup()
  
pilotB_data

# Calculate mean and standard error for total_reward
data_summary <- pilotB_data %>%
  group_by(transfer, teacher, game_sequence, block_index) %>%
  summarise(
    mean_reward = mean(total_reward),
    se_reward = sd(total_reward) / sqrt(n()),  # Standard error
    .groups = 'drop'  # This ensures the result is not grouped
  )

data_summary
```

```{r}
# Mapping block_index values
data_summary <- data_summary %>%
  mutate(block_index_mapped = case_when(
    game_sequence == "play-play-play" & block_index == 0 ~ 24,
    game_sequence == "play-play-play" & block_index == 1 ~ 48,
    game_sequence == "play-play-play" & block_index == 2 ~ 72,
    game_sequence != "play-play-play" & block_index == 0 ~ 24,
    game_sequence != "play-play-play" & block_index == 1 ~ 48,
    game_sequence != "play-play-play" & block_index == 2 ~ 52,
    game_sequence != "play-play-play" & block_index == 3 ~ 76,
    TRUE ~ NA_real_  # This line ensures that any other cases result in NA
  ))

# Correctly assigning factor levels and labels for game_sequence
data_summary$game_sequence <- factor(data_summary$game_sequence, 
                                     levels = c("play-play-play", 
                                                "play-watch-indicate-play", 
                                                "play-play-indicate-play"))

# Correctly assigning factor levels and labels for transfer with False first
data_summary$transfer <- factor(data_summary$transfer, levels = c(FALSE, TRUE), labels = c("Transfer: False", "Transfer: True"))

# Create the plot with the adjusted facet grid layout using the mapped block_index
p <- ggplot(data_summary, aes(x = block_index_mapped, y = mean_reward, group = teacher, color = teacher)) +
  geom_line() +  # Line for mean rewards
  geom_errorbar(aes(ymin = mean_reward - se_reward, ymax = mean_reward + se_reward), width = 0.1) +
  facet_grid(transfer ~ game_sequence, scales = "free_x") +  # Adjusted for Transfer on rows and Game Sequence on columns
  scale_color_manual(values = c("expert" = "blue", "novice" = "red", "none" = "green")) +
  scale_x_continuous(breaks = c(24, 48, 52, 76), labels = c("24", "48", "52", "76")) +  # Set custom x-axis ticks
  labs(title = "Game Reward Analysis (Mean and SE)",
       x = "Number of Trials (Shots)",
       y = "Mean Total Reward") +
  theme_minimal() +
  theme(strip.background = element_blank(), 
        strip.text.x = element_text(size = 12, face = "bold"), 
        strip.text.y = element_text(size = 12, face = "bold"),
        legend.position = "bottom",
        legend.title = element_blank())

# Save and display the plot
ggsave(here("figures", "transfer_pilotb.png"), plot = p, width = 7.5, height = 5)

# Display the plot
p
```

## Pilot Filter Data
```{r}
pilotC_demographics_raw <- read_csv(here("data","socialtransferbandits_pilotc_prolific_demographics.csv"))
pilotC_demographics_filtered <- pilotC_demographics_raw |>
  dplyr::filter(!(Status == "RETURNED" | Status == "TIMED-OUT" | grepl("^Yes", Colourblindness)))
pilotC_demographics_filtered

pilotC_data_raw <- read_csv(here("data","socialtransferbandits_pilotc_amg_gamelogs.csv")) 
pilotC_data_raw
pilotC_data_filtered <- pilotC_data_raw |>
  filter(prolific_id %in% pilotC_demographics_filtered$`Participant id`) |>
  filter(game_type != 'practice') |>
  filter(game_type != 'complete') |>
  filter(prolific_id != 'julio_test') 
pilotC_data_filtered
```

```{r}

```
## Mean Total Reward Plots
```{r}
# Calculate mean and standard error for total_reward
data_summary <- pilotC_data_filtered %>%
  group_by(prolific_id, session_index, session_id, game_type,
           game_sequence, teacher, transfer, block_index) |>
  summarise(total_reward = max(game_total_reward)) |>
  ungroup() |>
  group_by(transfer, teacher, game_sequence, block_index) |>
  summarise(
    mean_reward = mean(total_reward),
    se_reward = sd(total_reward) / sqrt(n()),  # Standard error
    .groups = 'drop'  # This ensures the result is not grouped
  )

data_summary

# Mapping block_index values
data_summary <- data_summary |>
  mutate(block_index_mapped = case_when(
    game_sequence == "play-play-play" & block_index == 0 ~ 24,
    game_sequence == "play-play-play" & block_index == 1 ~ 48,
    game_sequence == "play-play-play" & block_index == 2 ~ 72,
    game_sequence != "play-play-play" & block_index == 0 ~ 24,
    game_sequence != "play-play-play" & block_index == 1 ~ 48,
    game_sequence != "play-play-play" & block_index == 2 ~ 52,
    game_sequence != "play-play-play" & block_index == 3 ~ 76,
    TRUE ~ NA_real_  # This line ensures that any other cases result in NA
  ))

# Correctly assigning factor levels and labels for game_sequence
data_summary$game_sequence <- factor(data_summary$game_sequence, 
                                     levels = c("play-play-play", 
                                                "play-watch-indicate-play", 
                                                "play-play-indicate-play"))

# Correctly assigning factor levels and labels for transfer with False first
data_summary$transfer <- factor(data_summary$transfer, levels = c(FALSE, TRUE), labels = c("Transfer: False", "Transfer: True"))

# Create the plot with the adjusted facet grid layout using the mapped block_index
p <- ggplot(data_summary, aes(x = block_index_mapped, y = mean_reward, group = teacher, color = teacher)) +
  geom_line() +  # Line for mean rewards
  geom_errorbar(aes(ymin = mean_reward - se_reward, ymax = mean_reward + se_reward), width = 0.1) +
  facet_grid(transfer ~ game_sequence, scales = "free_x") +  # Adjusted for Transfer on rows and Game Sequence on columns
  scale_color_manual(values = c("expert" = "blue", "novice" = "red", "none" = "green")) +
  scale_x_continuous(breaks = c(24, 48, 52, 76), labels = c("24", "48", "52", "76")) +  # Set custom x-axis ticks
  labs(title = "Game Reward Analysis (Mean and SE)",
       x = "Number of Trials (Shots)",
       y = "Mean Total Reward") +
  theme_minimal() +
  theme(strip.background = element_blank(), 
        strip.text.x = element_text(size = 12, face = "bold"), 
        strip.text.y = element_text(size = 12, face = "bold"),
        legend.position = "bottom",
        legend.title = element_blank())

# Save and display the plot
ggsave(here("figures", "transfer_pilotc_reward.png"), plot = p, width = 7.5, height = 5, bg="white")

# Display the plot
p
```


## Mean Entropy Plots
```{r}
# Function to calculate entropy
calculate_entropy <- function(x) {
  # Calculate the proportion of each unique position
  p <- table(x) / length(x)
  # Calculate entropy
  entropy <- -sum(p * log2(p))
  return(entropy)
}

# Filter the data to include only rows where action is 'shoot' or 'indicate'
filtered_data <- pilotC_data_filtered %>%
  filter(action %in% c("shoot", "indicate"))

# Compute the entropy of position within each group
entropy_summary <- filtered_data %>%
  group_by(prolific_id, session_index, session_id, game_type,
           game_sequence, teacher, transfer, block_index) %>%
  summarise(position_entropy = calculate_entropy(position)) %>%
  ungroup()

# Calculate the mean of the position entropy and its standard error
mean_entropy_summary <- entropy_summary %>%
  group_by(transfer, teacher, game_sequence, block_index) %>%
  summarise(
    mean_position_entropy = mean(position_entropy, na.rm = TRUE),
    se_position_entropy = sd(position_entropy, na.rm = TRUE) / sqrt(n()),  # Standard error
    .groups = 'drop'
  )

mean_entropy_summary

# Mapping block_index values
mean_entropy_summary <- mean_entropy_summary %>%
  mutate(block_index_mapped = case_when(
    game_sequence == "play-play-play" & block_index == 0 ~ 24,
    game_sequence == "play-play-play" & block_index == 1 ~ 48,
    game_sequence == "play-play-play" & block_index == 2 ~ 72,
    game_sequence != "play-play-play" & block_index == 0 ~ 24,
    game_sequence != "play-play-play" & block_index == 1 ~ 48,
    game_sequence != "play-play-play" & block_index == 2 ~ 52,
    game_sequence != "play-play-play" & block_index == 3 ~ 76,
    TRUE ~ NA_real_  # This line ensures that any other cases result in NA
  ))

# Correctly assigning factor levels and labels for game_sequence
mean_entropy_summary$game_sequence <- factor(mean_entropy_summary$game_sequence, 
                                     levels = c("play-play-play", 
                                                "play-watch-indicate-play", 
                                                "play-play-indicate-play"))

# Correctly assigning factor levels and labels for transfer with False first
mean_entropy_summary$transfer <- factor(mean_entropy_summary$transfer, levels = c(FALSE, TRUE), labels = c("Transfer: False", "Transfer: True"))

# Create the plot with the adjusted facet grid layout using the mapped block_index
p <- ggplot(mean_entropy_summary, aes(x = block_index_mapped, y = mean_position_entropy, group = teacher, color = teacher)) +
  geom_line() +  # Line for mean rewards
  geom_errorbar(aes(ymin = mean_position_entropy - se_position_entropy, ymax = mean_position_entropy + se_position_entropy), width = 0.1) +
  facet_grid(transfer ~ game_sequence, scales = "free_x") +  # Adjusted for Transfer on rows and Game Sequence on columns
  scale_color_manual(values = c("expert" = "blue", "novice" = "red", "none" = "green")) +
  scale_x_continuous(breaks = c(24, 48, 52, 76), labels = c("24", "48", "52", "76")) +  # Set custom x-axis ticks
  labs(title = "Game Reward Analysis (Mean and SE)",
       x = "Number of Trials (Shots)",
       y = "Mean Position Entropy") +
  theme_minimal() +
  theme(strip.background = element_blank(), 
        strip.text.x = element_text(size = 12, face = "bold"), 
        strip.text.y = element_text(size = 12, face = "bold"),
        legend.position = "bottom",
        legend.title = element_blank())

# Save and display the plot
ggsave(here("figures", "transfer_pilotc_entropy.png"), plot = p, width = 7.5, height = 5, bg="white")

# Display the plot
p
```