Update stats_data.py

stats_data.py

@@ -1,6 +1,6 @@
 import pandas as pd
 
-# Metadata for speakers available in the demo
+# Speaker meta remains the same for UI lookups
 SPEAKER_META = {
     "M05": {"Gender": "Male", "Severity": "Severe", "Dataset": "Torgo"},
     "F01": {"Gender": "Female", "Severity": "Severe", "Dataset": "Torgo"},
@@ -14,42 +14,46 @@ SPEAKER_META = {
 }
 
 def get_indomain_breakdown():
-    # Primary Data for Torgo In-Domain (5K Model is the Accuracy Champion)
     data = {
         "Speaker": ["M05", "F01", "M01", "M04", "M02", "M03", "F03", "F04"],
         "Severity": ["Severe", "Severe", "Moderate", "Moderate", "Mild", "Mild", "Mild", "Mild"],
-        "Whisper Tiny": [12.1, 12.6, 32.7, 31.8, 62.1, 58.4, 61.2, 59.1],
-        "DSR Lab (5K)": [33.1, 34.2, 47.2, 45.6, 84.5, 81.8, 83.5, 82.8]
+        "Whisper Baseline": [12.1, 12.6, 32.7, 31.8, 62.1, 58.4, 61.2, 59.1],
+        "Correction Layer (5K)": [33.1, 34.2, 47.2, 45.6, 84.5, 81.8, 83.5, 82.8]
     }
     df = pd.DataFrame(data)
+    # Relative Gain calculation: ((Ours - Whisper) / Whisper) * 100
+    df["Relative Gain"] = (((df["Correction Layer (5K)"] - df["Whisper Baseline"]) / df["Whisper Baseline"]) * 100).round(1)
     
-    # Calculate Gains
-    df["Absolute Gain (%)"] = (df["DSR Lab (5K)"] - df["Whisper Tiny"]).round(2)
-    df["Relative Improvement (%)"] = (((df["DSR Lab (5K)"] - df["Whisper Tiny"]) / df["Whisper Tiny"]) * 100).round(1)
-    
-    # Format for display
-    df["Whisper Tiny"] = df["Whisper Tiny"].astype(str) + "%"
-    df["DSR Lab (5K)"] = df["DSR Lab (5K)"].astype(str) + "%"
-    df["Absolute Gain (%)"] = "+" + df["Absolute Gain (%)"].astype(str) + "%"
-    df["Relative Improvement (%)"] = "+" + df["Relative Improvement (%)"].astype(str) + "%"
+    # Add percentage signs for display
+    for col in ["Whisper Baseline", "Correction Layer (5K)"]:
+        df[col] = df[col].astype(str) + "%"
+    df["Relative Gain"] = "+" + df["Relative Gain"].astype(str) + "%"
     
     return df
 
 def get_experimental_summary():
-    # Comparing conditions (In-Domain, LOSO, Zero-Shot)
     data = {
-        "Condition": ["In-Domain (Seen Torgo)", "LOSO (Unseen Torgo F01)", "Zero-Shot (UA-Speech F02)"],
-        "Whisper Tiny": [41.50, 12.38, 4.33],
-        "Our Best Score": [58.77, 24.76, 6.19]
+        "Experiment Condition": ["In-Domain (Seen Torgo)", "LOSO (Unseen Torgo F01)", "Zero-Shot (UA-Speech F02)"],
+        "Whisper Baseline": [41.50, 12.38, 4.33],
+        "5K Model Score": [58.77, "N/A", 6.19],
+        "10K Model Score": [54.67, 24.76, 5.98]
     }
     df = pd.DataFrame(data)
+    # Calculate gain based on the best performing model in that row
+    best_scores = [58.77, 24.76, 6.19]
+    whisper_base = [41.50, 12.38, 4.33]
+    gains = []
+    for b, w in zip(best_scores, whisper_base):
+        gains.append(f"+{round(((b-w)/w)*100, 1)}%")
     
-    # Calculate Relative Gain
-    df["Relative Gain"] = (((df["Our Best Score"] - df["Whisper Tiny"]) / df["Whisper Tiny"]) * 100).round(1)
+    df["Relative Gain (Best)"] = gains
     
-    # Format for display
-    df["Whisper Tiny"] = df["Whisper Tiny"].astype(str) + "%"
-    df["Our Best Score"] = df["Our Best Score"].astype(str) + "%"
-    df["Relative Gain"] = "+" + df["Relative Gain"].astype(str) + "%"
+    # Formatting
+    df["Whisper Baseline"] = df["Whisper Baseline"].astype(str) + "%"
+    df.at[0, "5K Model Score"] = "58.77%"
+    df.at[2, "5K Model Score"] = "6.19%"
+    df.at[0, "10K Model Score"] = "54.67%"
+    df.at[1, "10K Model Score"] = "24.76%"
+    df.at[2, "10K Model Score"] = "5.98%"
     
     return df