app.py

import joblib
import pandas as pd
from flask import Flask, request, jsonify

# Initialize Flask app with a name
app = Flask("SuperKart sales prediction app backend")

# Load the trained churn prediction model
model = joblib.load("SuperKart_model_deployment_model_v1_0.joblib")

# Define a route for the home page
@app.get('/')
def home():
    return "Welcome to the SuperKart Sales Prediction API"

# Define an endpoint to predict sales of the single product in a store
@app.post('/v1/sales')
def predict_churn():
    # Get JSON data from the request
    store_data = request.get_json()

    # Extract relevant store features from the input data
    requestData = {
        'Product_Weight': store_data['Product_Weight'],
        'Product_Sugar_Content': store_data['Product_Sugar_Content'],
        'Product_Allocated_Area': store_data['Product_Allocated_Area'],
        'Product_Type': store_data['Product_Type'],
        'Product_MRP': store_data['Product_MRP'],
        'Store_Id': store_data['Store_Id'],
        'Store_Establishment_Year': store_data['Store_Establishment_Year'],
        'Store_Size': store_data['Store_Size'],
        'Store_Location_City_Type': store_data['Store_Location_City_Type'],
        'Store_Type': store_data['Store_Type'],
        'Product_Store_Sales_Total': store_data['Product_Store_Sales_Total']
    }

    # Convert the extracted data into a DataFrame
    input_data = pd.DataFrame([requestData])

    # create encoder with OneHotEncoder for encoding the selected values to match the training data
    encoder = OneHotEncoder(handle_unknown='ignore', sparse_output=False) # Important for handling unseen categories

    # You MUST use the *trained* encoder to transform the new data
    encoded_new_data = encoder.transform(input_data[['Product_Sugar_Content','Product_Type','Store_Id','Store_Size','Store_Location_City_Type','Store_Type']])

    # Make a churn prediction using the trained model
    prediction = model.predict(encoded_new_data).tolist()[0]

    #Calculate the actual price
    predicted_sales = np.exp(prediction)

    # Convert predicted_price to Python float
    predicted_sales = round(float(predicted_sales), 2)

    # Return the prediction as a JSON response
    return jsonify({'Predicted_Sale': predicted_sales})