app.py

import streamlit as st
import pandas as pd
import joblib
import numpy as np
from datetime import datetime

# -------------------------------
# Load the trained model safely
# -------------------------------
@st.cache_resource #is a Streamlit decorator used to cache expensive resources — things that take time to load or initialize
def load_model():
    # Use the path where the model was saved in the notebook
    saved_model_path = "deployment_files/car_prediction_model_v1_0.joblib"
    try:
        model = joblib.load(saved_model_path)
        return model
    except FileNotFoundError:
        st.error(f"Model file not found at {saved_model_path}. Please ensure the model is saved correctly.")
        return None
    except Exception as e:
        st.error(f"An error occurred while loading the model: {e}")
        return None


model = load_model()

# -------------------------------
# Main Streamlit App
# -------------------------------
def main():
    st.set_page_config(page_title="Car Pricing Predictor", page_icon="🚗", layout="centered")
    st.title("🚗 Car Pricing Prediction Solution")
    st.write("Enter your car details below to predict its resale price in **USD**.")

    # -------------------------------
    # Sidebar Instructions
    # -------------------------------
    st.sidebar.header("Instructions")
    st.sidebar.info("""
    - **Year:** Year the car was manufactured (e.g., 2016)
    - **Kms Driven:** Total kilometers driven
    - **Owner:** Number of previous owners (0 = first owner)
    - **Fuel Type:** Diesel, Petrol, or CNG
    - **Seller Type:** Individual or Dealer
    - **Transmission:** Manual or Automatic
    """)

    # -------------------------------
    # User Inputs
    # -------------------------------
    manufacture_year_input = st.number_input(
        "Year of Manufacture",
        min_value=1990, max_value=datetime.now().year, value=2015,
        help="Enter the year the car was made"
    )

    kms_driven_input = st.number_input(
        "Total Kilometers Driven",
        min_value=0, value=50000, step=1000,
        help="Enter total kilometers driven"
    )

    owner_input = st.number_input(
        "Number of Previous Owners",
        min_value=0, max_value=10, value=0, step=1,
        help="Enter the number of previous owners"
    )

    fuel_type_input = st.selectbox(
        "Fuel Type",
        ["Petrol", "Diesel", "CNG"],
        help="Select the fuel type"
    )

    seller_type_input = st.selectbox(
        "Seller Type",
        ["Dealer", "Individual"],
        help="Select the seller type"
    )

    transmission_input = st.selectbox(
        "Transmission Type",
        ["Manual", "Automatic"],
        help="Select the transmission type"
    )

    # -------------------------------
    # Prediction Logic
    # -------------------------------
    if st.button("Predict Selling Price"):
        if model is not None:
            # Create a DataFrame from user inputs
            # Ensure column order matches the training data used for the model
            # The model was trained on: ['Year', 'Kms_Driven', 'Owner', 'Fuel_Type_Diesel', 'Fuel_Type_Petrol', 'Seller_Type_Individual', 'Transmission_Manual']

            # Calculate Car Age
            current_year = datetime.now().year
            car_age = current_year - manufacture_year_input

            input_data = pd.DataFrame({
                'Year': [car_age], # Use calculated car age
                'Kms_Driven': [kms_driven_input],
                'Owner': [owner_input],
                # One-hot encode Fuel_Type
                'Fuel_Type_Diesel': [1 if fuel_type_input == 'Diesel' else 0],
                'Fuel_Type_Petrol': [1 if fuel_type_input == 'Petrol' else 0],
                # One-hot encode Seller_Type
                'Seller_Type_Individual': [1 if seller_type_input == 'Individual' else 0],
                # One-hot encode Transmission
                'Transmission_Manual': [1 if transmission_input == 'Manual' else 0]
            })

            # Ensure numerical columns have correct types and check for NaNs
            numerical_cols = ['Year', 'Kms_Driven', 'Owner']
            for col in numerical_cols:
                 input_data[col] = pd.to_numeric(input_data[col], errors='coerce')

            if input_data[numerical_cols].isnull().any().any():
                 st.error("Invalid input detected for numerical fields. Please check your values.")
                 return # Stop execution if invalid input is found


            # Ensure one-hot encoded columns have correct types (int)
            encoded_cols = ['Fuel_Type_Diesel', 'Fuel_Type_Petrol', 'Seller_Type_Individual', 'Transmission_Manual']
            for col in encoded_cols:
                 input_data[col] = input_data[col].astype(int)


            try:
                # Make prediction
                predicted_price = model.predict(input_data)[0]

                # Display prediction
                st.success(f"Predicted Selling Price: **${predicted_price:,.2f}**")

            except Exception as e:
                st.error(f"An error occurred during prediction: {e}")
                st.write("Please check your input values and ensure the model is loaded correctly.")
        else:
            st.warning("Model not loaded. Please check the model file path and try again.")