Update app.py

app.py

@@ -1,87 +1,46 @@
 from flask import Flask, request, jsonify
-import math
+from flask_cors import CORS
+import joblib
+import pandas as pd
+import os
 
 app = Flask(__name__)
+CORS(app)
 
-
-
-# Skill weights remain the same
-skill_weights = {
-    'java': 3, 'python': 3, 'react': 3, 'c++': 3, 'sql': 3, 'aws': 3, 'ai': 3, 'ml': 3,
-    'tensorflow': 3, 'nodejs': 3, 'swift': 3, 'ios': 3, 'cybersecurity': 3,
-    'devops': 3, 'docker': 3, 'go': 3, 'microservices': 3, 'qa': 3, 'selenium': 3,
-    'data science': 3, 'vhdl': 3, 'verilog': 3, 'embedded c': 3, 'rtos': 3,
-    'vlsi': 3, 'cad': 3, 'solidworks': 3, 'robotics': 3, 'staadpro': 3,
-    'html': 2, 'css': 2, 'javascript': 2, 'ui/ux': 2, 'figma': 2, 'autocad': 2,
-    'thermodynamics': 2, 'manufacturing': 2, 'signal processing': 2, 'analog design': 2,
-}
-
-# Branch match bonus also remains
-BRANCH_MATCH_BONUS = 10
-
-def calculate_alumni_score(viewer_profile, target_profile):
-    viewer_skills = set(str(viewer_profile.get('viewer_skills', '')).lower().split('|'))
-    target_skills = set(str(target_profile.get('target_skills', '')).lower().split('|'))
-    viewer_branch = str(viewer_profile.get('viewer_branch', '')).lower().strip()
-    target_branch = str(target_profile.get('target_branch', '')).lower().strip()
-
-    # 1. Calculate the raw score from common skills and branch bonus (same as before)
-    common_skills = viewer_skills.intersection(target_skills)
-    skill_score = sum(skill_weights.get(skill, 1) for skill in common_skills)
-    
-    branch_score = 0
-    if viewer_branch and target_branch and viewer_branch == target_branch:
-        branch_score = BRANCH_MATCH_BONUS
-        
-    base_score = skill_score + branch_score
-
-    # 2. **THE KEY IMPROVEMENT**: Calculate Jaccard Similarity for skills
-    # Jaccard Score = (Number of Shared Skills) / (Total Number of Unique Skills)
-    # This measures the *proportion* of similarity.
-    union_skills = viewer_skills.union(target_skills)
-    if not union_skills:
-        jaccard_similarity = 0
-    else:
-        jaccard_similarity = len(common_skills) / len(union_skills)
-
-    # 3. Combine the base score with the similarity score.
-    # The similarity acts as a powerful multiplier. A low similarity will
-    # heavily reduce the score, even if the base_score is high.
-    final_score = base_score * jaccard_similarity
-    
-    # 4. Normalize to the 1-10 scale for the UI
-    if final_score == 0:
-        return 0
-    
-    normalized_score = min(10, math.log(final_score + 1) * 3.0) # Adjusted multiplier for better spread
-    
-    return round(normalized_score, 1)
-
+# In Spaces, files are in the same directory
+model = joblib.load('alumni_match_model.joblib')
+model_columns = joblib.load('model_feature_columns.joblib')
 
 @app.route('/', methods=['POST'])
-def get_match_score():
-    try:
-        data = request.get_json()
-        
-        viewer_profile = {
-            'viewer_skills': data.get('viewer_skills'),
-            'viewer_branch': data.get('viewer_branch')
-        }
-        target_profile = {
-            'target_skills': data.get('target_skills'),
-            'target_branch': data.get('target_branch')
-        }
-
-        if not viewer_profile or not target_profile:
-            return jsonify({"error": "Missing profile data"}), 400
-            
-        final_score = calculate_alumni_score(viewer_profile, target_profile)
+def handler():
+    incoming_data = request.get_json()
+    df = pd.DataFrame([incoming_data])
 
-        return jsonify({"score": final_score})
+    def count_common_skills(row):
+        viewer_skills = set(str(row.get('viewer_skills', '')).lower().split('|'))
+        target_skills = set(str(row.get('target_skills', '')).lower().split('|'))
+        return len(viewer_skills.intersection(target_skills))
 
-    except Exception as e:
-        print(f"An error occurred: {e}")
-        return jsonify({"error": "An internal error occurred."}), 500
+    df['common_skills_count'] = df.apply(count_common_skills, axis=1)
+    df['branch_match'] = (df['viewer_branch'].str.lower() == df['target_branch'].str.lower()).astype(int)
 
-if __name__ == '__main__':
-    app.run(debug=True)
+    for col in model_columns:
+        if col.startswith('company_'):
+            df[col] = 0
+    
+    company_name = incoming_data.get('target_company', '')
+    if company_name:
+        company_col_name = f"company_{company_name}"
+        if company_col_name in df.columns:
+            df[company_col_name] = 1
+
+    final_df = df[model_columns]
+    prediction_proba = model.predict_proba(final_df)
+    match_probability = prediction_proba[0][1]
+    final_score = round(match_probability * 10)
+
+    return jsonify({'score': final_score})
+
+# ADD THESE FINAL TWO LINES TO START THE SERVER
+if __name__ == "__main__":
+    app.run(host="0.0.0.0", port=7860)