added count, beautified boxplot output

batch_analysis_analysis.py

@@ -1,19 +1,16 @@
 import pandas as pd
-import numpy as np
 import argparse
 import seaborn as sns
 import matplotlib.pyplot as plt
 
-
 def parse_arguments():
     parser = argparse.ArgumentParser(description='Analyze chain data from CSV file.')
     parser.add_argument('--input', '-i', required=True, help='Path to the input CSV file')
     return parser.parse_args()
 
-
 def main():
     args = parse_arguments()
-    
+
     # Load the CSV file from the input argument
     df = pd.read_csv(args.input)
 
@@ -21,56 +18,85 @@ def main():
     if 'experiment_name' not in df.columns:
         raise ValueError("Input CSV must contain 'experiment_name' column.")
     experiment_name = df['experiment_name'].iloc[0]
-    
+
     # Strip timestamp from experiment_name if it exists
     experiment_name = experiment_name.split('-')[0] if '-' in experiment_name else experiment_name
-    
+
     # Group data by chain
     chain_groups = df.groupby('chain')
-    
+
-    # For each chain, create a plot with four boxplots (mean, std, min, max)
+    # For each chain, create a figure with five subplots for boxplots (mean, std, min, max, count)
     for chain_name, chain_data in chain_groups:
-        # Create a figure for this chain
+        fig, axs = plt.subplots(1, 5, figsize=(18, 6), constrained_layout=True)
-        plt.figure(figsize=(12, 8))
+
-        
         # Normalize chain name for filename
         chain_name_fs = str(chain_name).replace('--> /', '-').replace('/', '_').replace(' ', '')
-        
+
         # Create a DataFrame with the columns we want to plot
-        plot_data = pd.DataFrame({
+        plot_data = chain_data[['mean', 'std', 'min', 'max', 'count']].copy()
-            'Mean': chain_data['mean'],
+        plot_data.columns = ['Mean', 'Std', 'Min', 'Max', 'Count']
-            'Std': chain_data['std'],
+
-            'Min': chain_data['min'],
+        # Make all plots have the same color palette
-            'Max': chain_data['max']
+        palette = sns.color_palette("husl", 4)
-        })
+        # Add a distinct color for the 'Count' plot, as it is a different metric
-        
+        colors = palette + ['lightcoral']
-        # Create boxplots
+
-        ax = sns.boxplot(data=plot_data, palette='Set3')
+        for idx, (col, color) in enumerate(zip(['Mean', 'Std', 'Min', 'Max', 'Count'], colors)):
-        
+            ax = axs[idx]
-        # Add individual data points
+
-        sns.stripplot(data=plot_data, color='black', alpha=0.5, size=4, jitter=True)
+            # Create boxplots
-        
+            sns.boxplot(data=plot_data[col], ax=ax, color=color, showfliers=True, width=0.4)
-        # Set labels and title
+
-        plt.title(f'Statistics for Chain: {chain_name}\nAcross {len(chain_data)} Experiment Runs\n{experiment_name}', fontsize=14)
+            # Add individual data points
-        plt.ylabel('Latency (ms)', fontsize=12)
+            sns.swarmplot(data=plot_data[col], ax=ax, color='black', size=3, alpha=0.6)
-        plt.xlabel('Statistic Type', fontsize=12)
+
-        
+            # Set labels and title
-        # Add grid for better readability
+            ax.set_title(f'{col} Distribution', fontsize=14, fontweight='bold')
-        plt.grid(axis='y', linestyle='--', alpha=0.7)
+            ax.set_xticks([]) # Remove x-ticks for clarity
-        
+            ax.set_xlabel('') # No x-label needed
-        # Tighten layout and save the figure
+            ax.set_ylabel('Latency (ms)' if col != 'Count' else 'Count', fontsize=12)
+
+            # Calculate statistics of the statistics
+            data_values = plot_data[col]
+            stats_text = (
+                f"Mean: {data_values.mean():.2f}\n"
+                f"Std: {data_values.std():.2f}\n"
+                f"Min: {data_values.min():.2f}\n"
+                f"Max: {data_values.max():.2f}"
+            )
+
+            # --- Place legend in the top right using axes fraction coordinates ---
+            ax.text(
+                0.95, 0.98,  # axes fraction: 95% right, 98% up
+                stats_text,
+                transform=ax.transAxes,
+                verticalalignment='top',
+                horizontalalignment='right',
+                fontsize=10,
+                bbox=dict(facecolor='white', alpha=0.9, boxstyle='round,pad=0.3', edgecolor='gray')
+            )
+
+            # Add grid for better readability
+            ax.grid(axis='y', linestyle='--', alpha=0.4)
+
+        # Set the overall title for the figure
+        plt.suptitle(
+            f'Statistics for Chain: {chain_name}\nAcross {len(chain_data)} Experiment Runs - {experiment_name}',
+            fontsize=18, fontweight='bold'
+        )
+
+        # Save the figure with a filename that includes the chain name
         plt.tight_layout()
         output_file = args.input.replace('.csv', f'_chain_{chain_name_fs}_analysis.png')
         plt.savefig(output_file, dpi=300)
         plt.close()
-        
+
-        # Also calculate and print summary statistics for this chain
+        # Print summary statistics for the chain
         summary = chain_data.describe()
         print(f"\nSummary for chain: {chain_name}")
-        print(summary[['mean', 'std', 'min', 'max']])
+        print(summary[['mean', 'std', 'min', 'max', 'count']])
-        
-    print(f"\nAnalysis complete. Plots saved with base name: {args.input.replace('.csv', '_chain_*_analysis.png')}")
 
+    print(f"\nAnalysis complete. Plots saved with base name: {args.input.replace('.csv', '_chain_*_analysis.png')}")
 
 if __name__ == "__main__":
     main()

batch_analyze.py

@@ -51,7 +51,7 @@ def main(base_dir, name_filter):
             pm.execute_notebook(
                 "./trace-analysis.ipynb",
                 os.path.join(current_artifact, "output", "trace-analysis.ipynb"),
-                log_output=True
+                log_output=False
             )
         except Exception as e:
             LOGGER.exception(e)

trace-analysis.ipynb

@@ -683,6 +683,7 @@
     "    std_latency = np.std(e2e_latencies)\n",
     "    min_latency = np.min(e2e_latencies)\n",
     "    max_latency = np.max(e2e_latencies)\n",
+    "    count_latencies = len(e2e_latencies)\n",
     "    ax.axvline(mean_latency, c=\"red\", linewidth=2)\n",
     "    _, max_ylim = ax.get_ylim()\n",
     "    # Create a multi-line string with all stats\n",
@@ -690,7 +691,8 @@
     "        f\"Mean: {mean_latency:.2f} ms\\n\"\n",
     "        f\"Std:  {std_latency:.2f} ms\\n\"\n",
     "        f\"Min:  {min_latency:.2f} ms\\n\"\n",
-    "        f\"Max:  {max_latency:.2f} ms\"\n",
+    "        f\"Max:  {max_latency:.2f} ms\\n\"\n",
+    "        f\"Count: {count_latencies}\"\n",
     "    )\n",
     "    # Place text near top right of plot\n",
     "    ax.text(\n",
@@ -703,10 +705,10 @@
     "        bbox=dict(facecolor='white', alpha=0.7, boxstyle='round,pad=0.3')\n",
     "    )\n",
     "    plt.savefig(os.path.join(OUT_PATH, f\"plot_e2es_{name}.png\"))\n",
-    "    result_strings.append(f\"Chain {topics[0]} --> {topics[-1]} E2E stats: Mean: {mean_latency:.2f} ms, Std: {std_latency:.2f} ms, Min: {min_latency:.2f} ms, Max: {max_latency:.2f} ms\")\n",
+    "    result_strings.append(f\"Chain {topics[0]} --> {topics[-1]} E2E stats: Mean: {mean_latency:.2f} ms, Std: {std_latency:.2f} ms, Min: {min_latency:.2f} ms, Max: {max_latency:.2f} ms, Count: {count_latencies}\")\n",
     "    # also do it as csv of order: exepriment_name, chain, mean, std, min, max\n",
     "    result_strings_csv.append(\n",
-    "        f\"{EXPERIMENT_NAME},{topics[0]} --> {topics[-1]},{mean_latency:.2f},{std_latency:.2f},{min_latency:.2f},{max_latency:.2f}\"\n",
+    "        f\"{EXPERIMENT_NAME},{topics[0]} --> {topics[-1]},{mean_latency:.2f},{std_latency:.2f},{min_latency:.2f},{max_latency:.2f},{count_latencies}\"\n",
     "    )\n",
     "\n",
     "    ##################################################\n",