Updates to figures

Author: micahgallen

code/analyze_hrd_data.R

@@ -15,26 +15,42 @@ source("code/trials2counts.R")
 source("code/fit_metad_indiv.R")
 source("code/calc_auroc2.R")
 
-analyze_hrd_data <- function(hrd_data, nRatings = 4, plot_results = TRUE, show_traceplot = TRUE) {
+analyze_hrd_data <- function(hrd_data, nRatings = 4, plot_results = TRUE, show_traceplot = TRUE, participant_id = "unknown") {
 
   # Ensure factors are correctly formatted
   hrd_data <- make_factors(hrd_data)
 
   # Process HRD data with the specified number of rating bins
   processed_hrd_data <- process_hrd_data(hrd_data, nRatings)
 
+  # Ensure figs/ directory exists
+  if (!dir.exists("figs")) {
+    dir.create("figs")
+  }
+  
   # Generate and display plots if requested
   if (plot_results) {
     confidence_hist <- plot_confidence_histogram(hrd_data)
     trial_alpha_plot <- plot_trial_alpha(hrd_data)
-    print(confidence_hist + trial_alpha_plot)
+    combined_plot <- confidence_hist + trial_alpha_plot
+    
+    print(combined_plot)
+    
+    # Save figure
+    ggsave(filename = sprintf("figs/%s_confidence_trial_alpha.png", participant_id), plot = combined_plot, width = 8, height = 4, dpi = 150)
   }
 
   # Extract inputs for trials2counts
   stimID <- processed_hrd_data$Signal
   response <- processed_hrd_data$Response
   rating <- processed_hrd_data$ConfidenceBinned
 
+  # Remove NA values before trials2counts
+  valid_trials <- complete.cases(stimID, response, rating)
+  stimID <- stimID[valid_trials]
+  response <- response[valid_trials]
+  rating <- rating[valid_trials]
+  
   # Convert trials to counts
   counts <- trials2counts(stimID, response, rating, nRatings)
   nR_S1 <- counts[[1]]
@@ -71,25 +87,23 @@ analyze_hrd_data <- function(hrd_data, nRatings = 4, plot_results = TRUE, show_t
   # Compute mean confidence
   mean_confidence <- mean(hrd_data$Confidence, na.rm = TRUE)
 
-  # Compute mean accuracy
-
-  # Ensure ResponseCorrect is numeric (convert from factor levels "FALSE"/"TRUE" to 0/1)
+  # Ensure ResponseCorrect is numeric
   hrd_data$ResponseCorrect <- as.numeric(hrd_data$ResponseCorrect) - 1
 
   # Compute mean accuracy
   mean_accuracy <- mean(hrd_data$ResponseCorrect, na.rm = TRUE)
 
-  
   # Extract final EstimatedThreshold and EstimatedSlope values
   estimated_threshold <- tail(hrd_data$EstimatedThreshold, 1)
   estimated_slope <- tail(hrd_data$EstimatedSlope, 1)
 
   # Print results
   message(sprintf("Metacognition scores: AUROC = %.2f, MRatio = %.2f, Mean Confidence = %.2f", auroc, mratio, mean_confidence))
 
-  # Show traceplot if requested
+  # Show traceplot if requested and save it
   if (show_traceplot) {
     traceplot(output)
+    ggsave(filename = sprintf("figs/%s_traceplot.png", participant_id), width = 8, height = 4, dpi = 300)
   }
 
   # Generate posterior distribution plot
@@ -106,23 +120,25 @@ analyze_hrd_data <- function(hrd_data, nRatings = 4, plot_results = TRUE, show_t
 
   if (plot_results) {
     print(post_plot)
+    ggsave(filename = sprintf("figs/%s_posterior_meta_d.png", participant_id), plot = post_plot, width = 6, height = 4, dpi = 150)
   }
 
   # Return results as a dataframe
   results_df <- data.frame(
     mean_confidence = mean_confidence,
+    estimated_threshold = estimated_threshold,
+    estimated_slope = estimated_slope,
+    mean_accuracy = mean_accuracy,
+    auroc = auroc,
     d = d1,
     metad = metad,
-    mratio = mratio,
-    auroc = auroc,
-    mean_accuracy = mean_accuracy,
-    estimated_threshold = estimated_threshold,
-    estimated_slope = estimated_slope
+    mratio = mratio
   )
 
   return(results_df)
 }
 
+
 # Example usage:
 # hrd_data <- read_delim("path/to/data.txt", delim = ",")
 # results <- analyze_hrd_data(hrd_data, nRatings = 4, plot_results = TRUE, show_traceplot = TRUE)

code/analyze_study.R

@@ -29,7 +29,7 @@ for (file in log_files) {
   hrd_data <- read_delim(file, delim = ",")
 
   # Apply the HRD analysis function
-  result <- analyze_hrd_data(hrd_data, nRatings = 4, plot_results = FALSE, show_traceplot = FALSE)
+  result <- analyze_hrd_data(hrd_data, nRatings = 4, plot_results = TRUE, show_traceplot = FALSE, participant_id = participant_id)
 
   # Add participant ID and session type
   result <- result %>%

code/helper_functions.R

@@ -7,7 +7,7 @@ library(tidyverse)
 #'
 #' @return A ggplot2 theme object.
 theme_custom <- function() {
-  theme_minimal(base_size = 14) +
+  theme_minimal(base_size = 10) +
     theme(
       plot.title = element_text(face = "bold", hjust = 0.5),
       axis.title = element_text(face = "bold"),

figs/SS9909_confidence_trial_alpha.png

@@ -1,3 +1,3 @@
 "participant_id","session_type","estimated_threshold","estimated_slope","mean_accuracy","mean_confidence","auroc","d","metad","mratio"
-"SS9909","PRE",-41.1313575764749,22.1337813305331,0.475,64.975,0.552656104380242,-0.0619316234553173,-0.0582300985384358,0.940232070300045
-"SS9909","POST",-24.3720289784148,15.6519614790519,0.4875,69.506329113924,0.489089184060721,0.642754952497923,0.182892710102267,0.284545003335246
+"SS9909","PRE",-41.1313575764749,22.1337813305331,0.475,64.975,0.552656104380242,-0.0619316234553173,-0.0107351866295183,0.173339338298851
+"SS9909","POST",-24.3720289784148,15.6519614790519,0.4875,69.506329113924,0.489089184060721,0.642754952497923,0.141847292936391,0.220686425495646