Deploy backend Flask app

bot_detector_api.py

@@ -13,7 +13,7 @@ app = Flask("Bot detector")
 def home():
     return "✅ Welcome to the Bot Prediction API!"
 
-# Load models and artifacts
+# Load models and utilities
 model = joblib.load("model.joblib")
 encoders = joblib.load("encoders.joblib")
 scaler = joblib.load("scaler.joblib")
@@ -58,24 +58,31 @@ def prepare_features(row_dict):
             df[col] = pd.to_numeric(df[col], errors='coerce').fillna(0)
 
     df_scaled = scaler.transform(df)
-    iso_score = if_model.decision_function(df_scaled)
-    svm_score = svm_model.decision_function(df_scaled)
 
-    iso_prob = float(1 - iso_scaler.transform(iso_score.reshape(-1, 1))[0][0])
-    svm_prob = float(1 - svm_scaler.transform(svm_score.reshape(-1, 1))[0][0])
+    iso_score = if_model.decision_function(df_scaled).reshape(-1, 1)
+    svm_score = svm_model.decision_function(df_scaled).reshape(-1, 1)
+
+    iso_prob = 1 - iso_scaler.transform(iso_score)[0][0]
+    svm_prob = 1 - svm_scaler.transform(svm_score)[0][0]
 
     df['iso_anomaly_prob'] = iso_prob
     df['svm_anomaly_prob'] = svm_prob
 
     return df[feature_names]
 
-def generate_shap_bot_attack_paragraph(index, shap_values, X, encoders=None, class_index=1, top_n=10):
-    shap_vals = shap_values.values[index, :, class_index]
-    base_val = shap_values.base_values[index, class_index]
+def generate_shap_explanation(index, shap_values, X, encoders=None, class_index=1, top_n=10):
+    if isinstance(shap_values, list):
+        shap_vals = shap_values[class_index][index]
+        base_val = explainer.expected_value[class_index]
+    else:
+        shap_vals = shap_values[index]
+        base_val = explainer.expected_value if np.isscalar(explainer.expected_value) else explainer.expected_value[class_index]
+
+    shap_scalar_vals = [float(s[0]) if isinstance(s, np.ndarray) else float(s) for s in shap_vals]
+
     x_vals = X.iloc[index]
     feature_names = X.columns
 
-    # Decode encoded values
     decoded_vals = {}
     for col in feature_names:
         val = x_vals[col]
@@ -87,51 +94,52 @@ def generate_shap_bot_attack_paragraph(index, shap_values, X, encoders=None, cla
         except:
             decoded_vals[col] = val
 
-    feature_contribs = list(zip(feature_names, decoded_vals.values(), shap_vals))
+    feature_contribs = list(zip(feature_names, decoded_vals.values(), shap_scalar_vals))
     feature_contribs = sorted(feature_contribs, key=lambda x: abs(x[2]), reverse=True)[:top_n]
 
     positive_impacts = []
     negative_impacts = []
 
     for fname, fval, sval in feature_contribs:
-        line = f"  - {fname:20} = {str(fval):<20} contributed {abs(sval):.4f}"
+        line = f"  - {fname:20} = {str(fval):<25} contributed {sval:+.4f}"
         if sval > 0:
             positive_impacts.append(line)
         elif sval < 0:
             negative_impacts.append(line)
 
-    final_pred_val = base_val + np.sum(shap_vals)
+    final_log_odds = base_val + sum(shap_scalar_vals)
 
-    explanation = f"\n==== SHAP Explanation for Bot Attack Classification ====\n"
-    explanation += f"Base model prediction for class 1 (Bot Attack): {base_val:.4f}\n"
-    explanation += f"Final model output for this instance        : {final_pred_val:.4f}\n\n"
+    explanation = f"==== SHAP Explanation for Bot Attack Classification ====\n\n"
+    explanation += f"Base value (log-odds for class 1) : {base_val:.4f}\n"
+    explanation += f"Predicted log-odds (class 1)      : {final_log_odds:.4f}\n\n"
 
     if positive_impacts:
-        explanation += "🔺 Factors that INCREASED the likelihood of Bot Attack:\n" + "\n".join(positive_impacts) + "\n\n"
+        explanation += "🔺 Factors that INCREASED Bot Likelihood:\n" + "\n".join(positive_impacts) + "\n\n"
 
     if negative_impacts:
-        explanation += "🔻 Factors that DECREASED the likelihood of Bot Attack:\n" + "\n".join(negative_impacts) + "\n\n"
+        explanation += "🔻 Factors that DECREASED Bot Likelihood:\n" + "\n".join(negative_impacts) + "\n\n"
 
-    explanation += "📝 These features collectively explain the model's classification of this session.\n"
-    return explanation
+    explanation += "📝 These features collectively explain the model's decision.\n"
+    return explanation, base_val, final_log_odds
 
 @app.post('/v1/predict')
 def predict():
     try:
         row = request.get_json()
         X = prepare_features(row)
-        probs = model.predict_proba(X)[0]
-        pred_label = model.classes_[np.argmax(probs)]
+        pred_prob = model.predict_proba(X)[0][1]
+        pred_label = int(pred_prob >= 0.5)
 
-        shap_values = explainer(X)
-        explanation = generate_shap_bot_attack_paragraph(0, shap_values, X, encoders)
+        shap_values = explainer.shap_values(X)
+        explanation, base_val, final_log_odds = generate_shap_explanation(0, shap_values, X, encoders)
 
         return jsonify({
-            "Prediction": "Bot Attack" if pred_label == 1 else "Legitimate",
+            "Prediction": "Bot Attack" if pred_label else "Legitimate",
+            "SHAP Base Value": round(base_val, 4),
+            "SHAP Predicted Value": round(final_log_odds, 4),
             "SHAP Explanation": explanation
         })
 
     except Exception as e:
         traceback.print_exc()
         return jsonify({"error": str(e)}), 500
-