Deploy backend Flask app

bot_detector_api.py

@@ -1,19 +1,18 @@
 import numpy as np
-import pandas as pd
 import joblib
+import pandas as pd
 import shap
 from flask import Flask, request, jsonify
 from urllib.parse import urlparse, parse_qs
 
-# Initialize Flask
+# Initialize Flask app
 app = Flask("Bot detector")
 
-# Health check
 @app.get('/')
 def home():
     return "✅ Welcome to the Bot Prediction API!"
 
-# Load models and utilities
+# Load models and artifacts
 model = joblib.load("model.joblib")
 encoders = joblib.load("encoders.joblib")
 scaler = joblib.load("scaler.joblib")
@@ -21,7 +20,7 @@ if_model = joblib.load("best_if_model.joblib")
 svm_model = joblib.load("best_svm_model.joblib")
 feature_names = joblib.load("feature_names.joblib")
 
-# Initialize SHAP explainer
+# SHAP explainer
 explainer = shap.TreeExplainer(model)
 
 def parse_url_params(url):
@@ -70,7 +69,6 @@ def prepare_features(row_dict):
     return df[feature_names]
 
 def generate_shap_bot_attack_paragraph(index, shap_values, X, encoders=None, class_index=1, top_n=10):
-    # Handle both single-output and multi-output SHAP format
     if isinstance(shap_values, list) and len(shap_values) > class_index:
         shap_vals = shap_values[class_index][index]
         base_val = explainer.expected_value[class_index]
@@ -81,16 +79,17 @@ def generate_shap_bot_attack_paragraph(index, shap_values, X, encoders=None, cla
     x_vals = X.iloc[index]
     feature_names = X.columns
 
-    # Decode readable values
     decoded_vals = {}
     for col in feature_names:
         val = x_vals[col]
         try:
             if encoders and col in encoders:
+                if isinstance(val, (np.ndarray, list)):
+                    val = val[0]
                 decoded_vals[col] = encoders[col].inverse_transform([int(val)])[0]
             else:
                 decoded_vals[col] = val
-        except:
+        except Exception:
             decoded_vals[col] = val
 
     feature_contribs = list(zip(feature_names, decoded_vals.values(), shap_vals))
@@ -98,6 +97,7 @@ def generate_shap_bot_attack_paragraph(index, shap_values, X, encoders=None, cla
 
     positive_impacts = []
     negative_impacts = []
+
     for fname, fval, sval in feature_contribs:
         line = f"  • {fname} = {fval} (contributed {abs(sval):.4f})"
         if sval > 0:
@@ -105,26 +105,20 @@ def generate_shap_bot_attack_paragraph(index, shap_values, X, encoders=None, cla
         elif sval < 0:
             negative_impacts.append(line)
 
-    def sigmoid(x):
-        return 1 / (1 + np.exp(-x))
-
+    def sigmoid(x): return 1 / (1 + np.exp(-x))
     final_log_odds = base_val + np.sum(shap_vals)
     final_pred_prob = sigmoid(final_log_odds)
 
-    # Build paragraph-style explanation
     paragraph = f"🧠 **SHAP Explanation for Anomaly Classification**\n\n"
-    paragraph += f"- The model starts with a baseline (SHAP base value) of **{base_val:.4f}**, representing the initial log-odds for class 1 (anomaly).\n"
-    paragraph += f"- After considering all features, the final predicted probability of this session being an anomaly is **{final_pred_prob:.4f}**.\n\n"
+    paragraph += f"- SHAP base value (log-odds for class 1): **{base_val:.4f}**\n"
+    paragraph += f"- Final predicted probability (Anomaly): **{final_pred_prob:.4f}**\n\n"
 
     if positive_impacts:
-        paragraph += "**🔺 Features that increased anomaly likelihood:**\n"
-        paragraph += "\n".join(positive_impacts) + "\n\n"
-
+        paragraph += "**🔺 Features that increased anomaly likelihood:**\n" + "\n".join(positive_impacts) + "\n\n"
     if negative_impacts:
-        paragraph += "**🔻 Features that decreased anomaly likelihood:**\n"
-        paragraph += "\n".join(negative_impacts) + "\n\n"
+        paragraph += "**🔻 Features that decreased anomaly likelihood:**\n" + "\n".join(negative_impacts) + "\n\n"
 
-    paragraph += "These top features collectively influenced the model’s final prediction."
+    paragraph += "These top features collectively influenced the model’s decision."
     return paragraph
 
 @app.post('/v1/predict')
@@ -136,14 +130,13 @@ def predict():
         pred_prob = float(pred_probs[0][1])
         pred_label = int(pred_prob >= 0.5)
 
-        # SHAP Explanation
         shap_values = explainer.shap_values(X)
         explanation = generate_shap_bot_attack_paragraph(0, shap_values, X, encoders)
 
         return jsonify({
             "Prediction": "Bot Attack" if pred_label else "Legitimate",
-            "Bot Probability": round(pred_prob, 4),
             "SHAP Explanation": explanation
         })
+
     except Exception as e:
         return jsonify({"error": str(e)}), 500