Upload 4 files

Dockerfile

@@ -0,0 +1,28 @@
+FROM python:3.11-slim
+
+# Install system dependencies in a single layer
+RUN apt-get update \
+    && apt-get install -y build-essential libpq-dev \
+    && rm -rf /var/lib/apt/lists/*
+
+# Create a non-root user
+RUN useradd -m -u 1000 user
+
+# Set the working directory
+WORKDIR /app
+
+# Copy requirements first to leverage caching
+COPY ./requirements.txt requirements.txt
+RUN pip install --no-cache-dir --upgrade -r requirements.txt
+
+# Copy the rest of the application files (including model)
+COPY --chown=user:user . /app
+
+# Switch to non-root user
+USER user
+
+# Expose Hugging Face Spaces port
+EXPOSE 7860
+
+# Start the Flask app
+CMD ["python", "app.py"]

app.py

@@ -0,0 +1,92 @@
+# -*- coding: utf-8 -*-
+"""Backendapp.ipynb
+
+Automatically generated by Colab.
+
+Original file is located at
+    https://colab.research.google.com/drive/1PjbKz6_4cQUuHck90e_e2dJXA7oWpc9B
+"""
+
+# -*- coding: utf-8 -*-
+
+import os
+import pandas as pd
+import numpy as np
+import joblib
+from flask import Flask, request, jsonify
+
+# --- Define Constants and Global Model ---
+
+# Model File Name (Must match the file created in Rubric 5)
+MODEL_FILE = 'final_superkart_sales_forecaster.joblib'
+
+# The raw features expected by the pipeline. This order is CRITICAL.
+FEATURE_COLUMNS = [
+    'Product_Weight', 'Product_Sugar_Content', 'Product_Allocated_Area',
+    'Product_Type', 'Product_MRP', 'Store_Size', 'Store_Location_City_Type',
+    'Store_Type', 'Store_Establishment_Year', 'Product_Id'
+]
+
+# Load the model once when the server starts
+model = None
+try:
+    # Load the best model pipeline (which includes the preprocessor and the Tuned XGBoost)
+    model = joblib.load(MODEL_FILE)
+    print(f"Model loaded successfully from {MODEL_FILE}")
+except FileNotFoundError:
+    print(f"ERROR: Model file {MODEL_FILE} not found. Ensure it is in the deployment folder.")
+except Exception as e:
+    # This error typically happens due to mismatched joblib/scikit-learn/xgboost versions.
+    print(f"CRITICAL ERROR: Failed to load model. Check library versions. Error: {e}")
+
+# --- Flask Application Setup ---
+
+app = Flask(__name__)
+
+@app.route('/predict', methods=['POST'])
+def predict():
+    """
+    Receives JSON data for prediction, processes it, and returns the sales forecast.
+    Expected input: A JSON array of objects, where each object contains all FEATURE_COLUMNS.
+    """
+    if model is None:
+        return jsonify({"error": "Model is not loaded. Cannot perform prediction."}), 503
+
+    try:
+        # Get JSON data from the request
+        data = request.get_json(force=True)
+
+        # 1. Convert JSON list of dictionaries to DataFrame
+        df = pd.DataFrame(data)
+
+        # 2. Reorder columns to match the trained pipeline's expected input order
+        # This is CRITICAL because the ColumnTransformer relies on input order.
+        df = df[FEATURE_COLUMNS]
+
+        # 3. Predict on the log scale (pipeline handles all preprocessing)
+        log_predictions = model.predict(df)
+
+        # 4. Inverse transform predictions to the original sales scale
+        # np.expm1 is the inverse of np.log1p
+        predictions = np.expm1(log_predictions).tolist()
+
+        return jsonify({"predictions": predictions})
+
+    except KeyError as e:
+        # Handle cases where required features are missing
+        return jsonify({"error": f"Missing required feature: {e}. All features in FEATURE_COLUMNS must be present."}), 400
+    except Exception as e:
+        # Catch unexpected errors during prediction
+        return jsonify({"error": f"An unexpected error occurred during prediction: {e}"}), 500
+
+@app.route('/health', methods=['GET'])
+def health_check():
+    """A simple health check endpoint to verify the API and model are running."""
+    # This endpoint is what Hugging Face checks for to confirm the app is live.
+    status = "OK" if model is not None else "ERROR: Model not loaded"
+    return jsonify({"status": status, "model_ready": model is not None})
+
+# Standard run command for Docker/Hugging Face compatibility
+if __name__ == '__main__':
+    # Use 0.0.0.0 for Docker compatibility. We MUST use port 7860 for Hugging Face Spaces.
+    app.run(debug=False, host='0.0.0.0', port=7860)

final_superkart_sales_forecaster.joblib

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:77d5eb12ff255ae1d4c22f8efef44724894e450fd3065a43704c879d5ad4d0df
+size 184720

requirements.txt

@@ -0,0 +1,11 @@
+# Flask and core utilities
+flask==3.0.3
+joblib==1.4.0
+
+# Numerical and Data Handling
+numpy==1.26.4
+pandas==2.2.2
+
+# Machine Learning Libraries (exact versions used in training)
+scikit-learn==1.6.1
+xgboost==2.1.4