Upload app.py with huggingface_hub

app.py

@@ -1,42 +1,53 @@
-
 from flask import Flask, request, jsonify
 import joblib
 import pandas as pd
-import numpy as np
+import os
 
 app = Flask(__name__)
 
-# Load the serialized model bundle
-BUNDLE_FILENAME = 'best_model_random_forest.joblib'
-bundle = joblib.load(BUNDLE_FILENAME)
-model = bundle['model']
-feature_cols = bundle['feature_cols']
+MODEL_PATH = os.getenv("MODEL_PATH", "best_model_random_forest.joblib")
+
+# Load the model bundle safely
+if not os.path.exists(MODEL_PATH):
+    raise FileNotFoundError(f"Model file not found: {MODEL_PATH}. Please ensure it's uploaded to the Space.")
+
+bundle = joblib.load(MODEL_PATH)
 
+# Extract actual model from dict or use as-is
+if isinstance(bundle, dict):
+    model = bundle.get("model", None)
+    feature_cols = bundle.get("feature_cols", None)
+else:
+    model = bundle
+    feature_cols = None
 
-@app.route('/')
-def home():
-    return "Sales Forecasting Backend is running!"
+if model is None:
+    raise ValueError("Model object not found inside the loaded bundle. Please verify the saved file structure.")
 
-@app.route('/predict', methods=['POST'])
+@app.route("/health", methods=["GET"])
+def health():
+    return jsonify({
+        "status": "ok",
+        "model_path": MODEL_PATH
+    })
+
+@app.route("/predict", methods=["POST"])
 def predict():
     try:
         data = request.get_json(force=True)
-        # Convert the incoming data to a pandas DataFrame
-        # Assuming the incoming data is a list of dictionaries, where each dictionary is a data point
-        input_data = pd.DataFrame(data)
-
-        # Align columns with the training data, adding missing columns with a default value (e.g., 0 or NaN)
-        input_data_processed = input_data.reindex(columns=feature_cols, fill_value=0)
-
+        df = pd.DataFrame([data]) if isinstance(data, dict) else pd.DataFrame(data)
 
-        # Make predictions
-        predictions = model.predict(input_data_processed)
+        # Align columns with training data if available
+        if feature_cols is not None:
+            df = df.reindex(columns=feature_cols, fill_value=0)
 
-        # Return predictions as a JSON response
-        return jsonify(predictions.tolist())
+        preds = model.predict(df)
+        return jsonify({"predictions": [float(p) for p in preds]})
     except Exception as e:
-        return jsonify({'error': str(e)})
+        return jsonify({
+            "error": "Prediction failed",
+            "details": str(e)
+        })
 
-if __name__ == '__main__':
-    # Running on 0.0.0.0 makes it accessible externally, useful for deployment
-    app.run(host='0.0.0.0', port=5000)
+if __name__ == "__main__":
+    app.run(host="0.0.0.0", port=5000)