import pandas as pd import matplotlib.pyplot as plt import seaborn as sns import os import numpy as np from sklearn.cluster import DBSCAN def moving_average(data, window_size): """Compute the moving average of data.""" return data.rolling(window=window_size).mean() def identify_fixations(point_of_gaze_x, point_of_gaze_y, proximity_threshold=100, min_samples=2): # Combine X and Y coordinates into a 2D array coordinates = np.column_stack((point_of_gaze_x, point_of_gaze_y)) # Use DBSCAN to cluster points of gaze dbscan = DBSCAN(eps=proximity_threshold, min_samples=min_samples) cluster_labels = dbscan.fit_predict(coordinates) # Initialize list to store fixations fixations = [] # Iterate through unique cluster labels unique_labels = np.unique(cluster_labels) for label in unique_labels: if label == -1: # Ignore noise points (not assigned to any cluster) continue # Extract points belonging to the current cluster cluster_points = coordinates[cluster_labels == label] # Calculate the centroid of the cluster centroid_x = np.mean(cluster_points[:, 0]) centroid_y = np.mean(cluster_points[:, 1]) # Add the centroid to fixations list fixations.append([centroid_x, centroid_y]) return np.array(fixations) def generate_diagrams(user_id): # Generate the file path based on user ID csv_file = f"user_{user_id}_eye_tracking.csv" # Check if the file exists if not os.path.exists(csv_file): print(f"Error: File {csv_file} not found.") return None # Read the CSV file into a pandas DataFrame df = pd.read_csv(csv_file) # Adjust the duration to 30 seconds df = df[df["Time(s)"] <= 30] # Group the data by time and select the first Point of Gaze point for each second df_grouped = df.groupby(df["Time(s)"].astype(int)) # Extract the Point of Gaze coordinates and corresponding time point_of_gaze_x = [] point_of_gaze_y = [] seconds = [] window_ids = [] # New list to store window IDs corneal_reflections = [] # New list to store corneal reflections for second, data in df_grouped: point_of_gaze_x.append(data["Point of Gaze X"].iloc[0]) point_of_gaze_y.append(data["Point of Gaze Y"].iloc[0]) seconds.append(second) window_ids.append(data["Window ID"].iloc[0]) # Add window ID data corneal_reflections.append(data["Corneal Reflection"].iloc[0]) # Add corneal reflection data # Create subplots in a 3x3 grid with adjusted height ratios fig, axes = plt.subplots(3, 3, figsize=(15, 18), gridspec_kw={'height_ratios': [1, 1, 1.2]}) # Adjust the top space to make room for the titles of upper graphs plt.subplots_adjust(top=0.88) # Plot Point of Gaze coordinates axes[0, 0].scatter(point_of_gaze_x, point_of_gaze_y, c='blue', label='Point of Gaze') for i, txt in enumerate(seconds): axes[0, 0].annotate(txt, (point_of_gaze_x[i], point_of_gaze_y[i]), textcoords="offset points", xytext=(0,10), ha='center') axes[0, 0].set_xlabel("X-coordinate (pixels)") axes[0, 0].set_ylabel("Y-coordinate (pixels)") axes[0, 0].set_title("Point of Gaze Coordinates Over Time (Every Second)") axes[0, 0].legend() axes[0, 0].grid(True) # Plot saccadic movements for i in range(len(point_of_gaze_x) - 1): axes[0, 1].plot([point_of_gaze_x[i], point_of_gaze_x[i+1]], [point_of_gaze_y[i], point_of_gaze_y[i+1]], c='red') axes[0, 1].scatter(point_of_gaze_x, point_of_gaze_y, c='blue', label='Point of Gaze') axes[0, 1].set_xlabel("X-coordinate (pixels)") axes[0, 1].set_ylabel("Y-coordinate (pixels)") axes[0, 1].set_title("Saccadic Movements") axes[0, 1].legend() axes[0, 1].grid(True) # Plot corneal reflections over time axes[1, 2].plot(df["Time(s)"], df["Corneal Reflection"], c='purple') axes[1, 2].set_xlabel("Time (seconds)") axes[1, 2].set_ylabel("Corneal Reflection") axes[1, 2].set_title("Corneal Reflection Over Time") axes[1, 2].grid(True) # Plot movement pursuit for i in range(len(seconds) - 1): # Get start and end points x0, y0 = point_of_gaze_x[i], point_of_gaze_y[i] x1, y1 = point_of_gaze_x[i+1], point_of_gaze_y[i+1] # Number of bends num_bends = np.random.randint(3, 10) # Random number of bends between 3 and 10 bend_factors = np.random.rand(num_bends) * 0.2 # Random bend factors bend_points_x = np.linspace(x0, x1, num_bends + 2)[1:-1] # Bend points along X-axis bend_points_y = np.linspace(y0, y1, num_bends + 2)[1:-1] # Bend points along Y-axis # Apply random bend factors bend_points_x += (np.random.choice([-1, 1], size=num_bends) * bend_factors * (x1 - x0)) bend_points_y += (np.random.choice([-1, 1], size=num_bends) * bend_factors * (y1 - y0)) # Combine all points all_points_x = np.concatenate([[x0], bend_points_x, [x1]]) all_points_y = np.concatenate([[y0], bend_points_y, [y1]]) # Plot line segments for j in range(len(all_points_x) - 1): axes[0, 2].plot([all_points_x[j], all_points_x[j+1]], [all_points_y[j], all_points_y[j+1]], c='green', alpha=0.5) axes[0, 2].scatter(point_of_gaze_x, point_of_gaze_y, c='blue', label='Point of Gaze') axes[0, 2].set_xlabel("X-coordinate (pixels)") axes[0, 2].set_ylabel("Y-coordinate (pixels)") axes[0, 2].set_title("Movement Pursuit") axes[0, 2].legend() axes[0, 2].grid(True) # Plot pupil dilation over time window_size = 10 # Adjust the window size for smoothing smoothed_pupil_dilation = moving_average(df["Pupil Dilation"], window_size) axes[1, 0].plot(df["Time(s)"], smoothed_pupil_dilation, c='orange') axes[1, 0].set_xlabel("Time (seconds)") axes[1, 0].set_ylabel("Pupil Dilation") axes[1, 0].set_title(f"Smoothed Pupil Dilation (Moving Average, Window Size = {window_size})") axes[1, 0].grid(True) # Plot a heatmap of Point of Gaze locations using the "plasma" colormap for clarity sns.kdeplot(x=df["Point of Gaze X"], y=df["Point of Gaze Y"], cmap="plasma", fill=True, thresh=0, levels=100, clip=[(0, 1920), (0, 1080)], ax=axes[1, 1], alpha=0.8, cbar=True, cbar_kws={'label': 'Density'}) # Adjust the brightness of the heatmap axes[1, 1].collections[0].set_alpha(0.4) # Adjust alpha value to make bright spots brighter axes[1, 1].set_xlabel("X-coordinate (pixels)") axes[1, 1].set_ylabel("Y-coordinate (pixels)") axes[1, 1].set_title("Heatmap of Point of Gaze Locations") # Plot window IDs in a table table_ax = axes[2, 0].table(cellText=[[i, j] for i, j in zip(seconds[::10], window_ids[::10])], loc='center') table_ax.auto_set_font_size(False) table_ax.set_fontsize(10) table_ax.scale(1, 1.5) # Add a title to the table subplot with lowered position table_ax_title = axes[2, 0].text(0.5, 0.95, 'Window ID per 10 Seconds', fontsize=12, ha='center') # Remove axis for the window ID subplot axes[2, 0].axis('off') axes[2, 2].remove() # Plot fixations with clusters axes[2, 1].scatter(point_of_gaze_x, point_of_gaze_y, c='blue', label='Point of Gaze') # Identify fixations fixations = identify_fixations(point_of_gaze_x, point_of_gaze_y, proximity_threshold=100) # Plot fixations with larger circles for fixation in fixations: x, y = fixation radius = 50 # Adjust the radius of the circle as needed circle = plt.Circle((x, y), radius, color='red', fill=False) axes[2, 1].add_artist(circle) axes[2, 1].set_xlabel("X-coordinate (pixels)") axes[2, 1].set_ylabel("Y-coordinate (pixels)") axes[2, 1].set_title("Fixations with Clusters") axes[2, 1].legend() axes[2, 1].grid(True) # Adjust spacing between subplots plt.tight_layout() # Adjust spacing between subplots to avoid overlap of titles plt.subplots_adjust(hspace=0.35) return fig # Define NUM_USERS NUM_USERS = 3 # Call the function for each user's CSV file all_figures = [] for user_id in range(1, NUM_USERS + 1): fig = generate_diagrams(user_id) if fig is not None: all_figures.append(fig) # Display all figures together plt.show()