app.py

import streamlit as st
import pandas as pd
from huggingface_hub import hf_hub_download
import joblib

# Download and load the best engine model from Hugging Face Model Hub.
# The model is a scikit-learn compatible estimator saved with joblib.
model_path = hf_hub_download(
    repo_id='Garg06/Predictive-Maintenance-Model',
    filename='best_engine_model.joblib'
)
model = joblib.load(model_path)

# ── App header ────────────────────────────────────────────────────────────
st.title('Engine Predictive Maintenance App')
st.write("""
This application predicts whether an engine is **Faulty** or **Normal** based on
real-time sensor readings. Enter the current sensor values below to get a prediction.
""")

# ── Sensor input fields ───────────────────────────────────────────────────
st.header('Engine Sensor Readings')

# Engine RPM: observed range 61–2239, typical idle around 750–800 RPM.
engine_rpm = st.number_input(
    'Engine RPM',
    min_value=0.0, max_value=2500.0, value=800.0, step=1.0,
    help='Engine revolutions per minute. Typical idle: ~750 RPM.'
)

# Lub oil pressure: observed range 0–7.27 bar.
lub_oil_pressure = st.number_input(
    'Lub Oil Pressure (bar)',
    min_value=0.0, max_value=10.0, value=3.3, step=0.01,
    help='Lubrication oil pressure in bar. Healthy range: 2.5–4.5 bar.'
)

# Fuel pressure: observed range 0–21.14 bar.
fuel_pressure = st.number_input(
    'Fuel Pressure (bar)',
    min_value=0.0, max_value=25.0, value=6.7, step=0.01,
    help='Fuel system pressure in bar. Healthy range: 4.9–7.7 bar.'
)

# Coolant pressure: observed range 0–7.48 bar.
coolant_pressure = st.number_input(
    'Coolant Pressure (bar)',
    min_value=0.0, max_value=10.0, value=2.3, step=0.01,
    help='Engine coolant system pressure in bar. Healthy range: 1.6–2.8 bar.'
)

# Lub oil temperature: observed range 71–90 °C.
lub_oil_temp = st.number_input(
    'Lub Oil Temperature (°C)',
    min_value=60.0, max_value=100.0, value=77.6, step=0.1,
    help='Lubrication oil temperature in °C. Normal operating range: 75–80 °C.'
)

# Coolant temperature: observed range 62–196 °C.
coolant_temp = st.number_input(
    'Coolant Temperature (°C)',
    min_value=60.0, max_value=200.0, value=78.4, step=0.1,
    help='Engine coolant temperature in °C. Normal range: 74–83 °C.'
)

# ── Assemble raw sensor input ─────────────────────────────────────────────
# Column names must exactly match those used during model training.
input_data = pd.DataFrame([{
    'Engine rpm':       engine_rpm,
    'Lub oil pressure': lub_oil_pressure,
    'Fuel pressure':    fuel_pressure,
    'Coolant pressure': coolant_pressure,
    'lub oil temp':     lub_oil_temp,
    'Coolant temp':     coolant_temp,
}])

# ── Feature engineering — must match prep.py exactly ─────────────────────
# The model was trained on 12 features (6 raw + 6 engineered).
# The saved Pipeline handles StandardScaling internally.
eps = 1e-6
input_data['pressure_ratio_fuel_coolant'] = (
    input_data['Fuel pressure'] / (input_data['Coolant pressure'] + eps)
)
input_data['temp_delta_oil_coolant'] = (
    input_data['lub oil temp'] - input_data['Coolant temp']
)
input_data['lub_pressure_per_rpm'] = (
    (input_data['Lub oil pressure'] / (input_data['Engine rpm'] + eps)) * 1000
)
input_data['coolant_pressure_per_rpm'] = (
    (input_data['Coolant pressure'] / (input_data['Engine rpm'] + eps)) * 1000
)
input_data['thermal_stress_index'] = (
    (input_data['lub oil temp'] + input_data['Coolant temp']) / 2
)
input_data['pressure_diff_fuel_lub'] = (
    input_data['Fuel pressure'] - input_data['Lub oil pressure']
)

# ── Prediction ────────────────────────────────────────────────────────────
if st.button('Predict Engine Condition'):
    prediction = model.predict(input_data)[0]
    probability = model.predict_proba(input_data)[0]

    if prediction == 1:
        st.error(
            f'⚠️ **Faulty Engine Detected!**  '
            f'Confidence: {probability[1]:.1%}  '
            f'\n\nImmediate inspection is recommended to prevent breakdown.'
        )
    else:
        st.success(
            f'✅ **Engine is Normal.**  '
            f'Confidence: {probability[0]:.1%}  '
            f'\n\nAll sensor readings are within acceptable limits.'
        )

    st.subheader('Input Summary')
    st.dataframe(input_data)