Fix deployment: create missing Flask app, save preprocessor, improve KMeans efficiency

app/main.py

@@ -0,0 +1,63 @@
+import os
+import io
+import joblib
+import pandas as pd
+from flask import Flask, request, jsonify
+
+from src.feature_engineering import engineer_features
+from src.data_preprocessing import preprocess_data
+
+app = Flask(__name__)
+
+MODEL_PATH = os.path.join("models", "clustering_model.pkl")
+PREPROCESSOR_PATH = os.path.join("models", "preprocessor.pkl")
+
+# Load model and preprocessor once at startup (lazy on first request)
+_model = None
+_preprocessor = None
+
+
+def _load_artifacts():
+    global _model, _preprocessor
+    if _model is None:
+        _model = joblib.load(MODEL_PATH)
+    if _preprocessor is None:
+        _preprocessor = joblib.load(PREPROCESSOR_PATH)
+
+
+@app.route("/")
+def index():
+    return jsonify({"status": "ok", "message": "User Profiling & Segmentation API"})
+
+
+@app.route("/predict", methods=["POST"])
+def predict():
+    """
+    Accept a CSV file upload and return cluster segment labels.
+    Usage: POST /predict  with form-data field 'file' containing a CSV.
+    """
+    if "file" not in request.files:
+        return jsonify({"error": "No file part in the request"}), 400
+
+    file = request.files["file"]
+    if file.filename == "":
+        return jsonify({"error": "No file selected"}), 400
+
+    try:
+        df = pd.read_csv(io.StringIO(file.stream.read().decode("utf-8")))
+    except Exception as e:
+        return jsonify({"error": f"Failed to parse CSV: {str(e)}"}), 400
+
+    try:
+        _load_artifacts()
+        df = engineer_features(df)
+        X = _preprocessor.transform(df)
+        labels = _model.predict(X)
+        df["segment"] = labels
+        return jsonify({"segments": df["segment"].tolist()})
+    except Exception as e:
+        return jsonify({"error": str(e)}), 500
+
+
+if __name__ == "__main__":
+    app.run(debug=False, host="0.0.0.0", port=5000)

generate_model.py

@@ -4,6 +4,7 @@
 from src.feature_engineering import engineer_features
 from src.clustering import cluster_users
 import os
+import joblib
 
 # Load dataset
 df = pd.read_csv("data/user_profiles_for_ads.csv")
@@ -12,11 +13,14 @@
 df = engineer_features(df)
 
 # Preprocess features
-X, _ = preprocess_data(df)
+X, preprocessor = preprocess_data(df)
 
 # Make sure models folder exists
 os.makedirs("models", exist_ok=True)
 
+# Save the preprocessor for use during predictions
+joblib.dump(preprocessor, "models/preprocessor.pkl")
+
 # Perform clustering and save model
 labels = cluster_users(X, k=4)
 
@@ -26,4 +30,5 @@
 df.to_csv("outputs/segmented_users.csv", index=False)
 
 print("✅ clustering_model.pkl saved to 'models/'")
+print("✅ preprocessor.pkl saved to 'models/'")
 print("✅ Segmented users saved to 'outputs/segmented_users.csv'")

run_app.sh

@@ -1,3 +1,2 @@
 #!/bin/bash
-export FLASK_APP=app/main.py
-flask run
+python app/main.py

src/clustering.py

@@ -3,7 +3,7 @@
 import joblib
 
 def cluster_users(X, k=4):
-    kmeans = KMeans(n_clusters=k, random_state=42)
+    kmeans = KMeans(n_clusters=k, random_state=42, n_init=10)
     labels = kmeans.fit_predict(X)
     joblib.dump(kmeans, 'models/clustering_model.pkl')
     return labels