app.py

from flask import Flask, render_template, request, jsonify
import numpy as np
import tensorflow as tf
import joblib

app = Flask(__name__)

# --- MODEL LOADING ---
# Load the model and scaler you want to use.
# By default, this loads the FIRST model you trained.
try:
    model = tf.keras.models.load_model('water_quality_model.h5')
    scaler = joblib.load('scaler.joblib')
    print("✅ Model 'water_quality_model.h5' and 'scaler.joblib' loaded successfully.")

except Exception as e:
    print(f"❌ Error loading 'water_quality_model.h5': {e}")
    print("---")
    # --- IF YOU WANT TO USE THE ALTERNATIVE MODEL, UNCOMMENT THE 3 LINES BELOW ---
    # try:
    #     model = tf.keras.models.load_model('water_quality_model_ALT.h5')
    #     scaler = joblib.load('scaler_ALT.joblib')
    #     print("✅ Alternative Model 'water_quality_model_ALT.h5' loaded.")
    # except Exception as e2:
    #     print(f"❌ Error loading all models: {e2}")
    #     model = None
    #     scaler = None
    model = None
    scaler = None
# ---------------------


@app.route('/')
def home():
    """Renders the main HTML page."""
    return render_template('index.html')


@app.route('/predict', methods=['POST'])
def predict():
    """Receives data from the form and returns a prediction."""
    if not model or not scaler:
        return jsonify({'error': 'Model is not loaded properly. Check server logs.'}), 500

    try:
        # Get data from the JSON request sent by the JavaScript
        data = request.get_json()
        
        # 1. Create a numpy array in the correct order (must match training order)
        features = [
            float(data['ph']),
            float(data['Hardness']),
            float(data['Solids']),
            float(data['Chloramines']),
            float(data['Sulfate']),
            float(data['Conductivity']),
            float(data['Organic_carbon']),
            float(data['Trihalomethanes']),
            float(data['Turbidity'])
        ]
        
        # 2. Put data into a 2D numpy array (model expects a batch)
        input_data = np.array([features])
        
        # 3. Scale the data using the *loaded* scaler
        input_scaled = scaler.transform(input_data)
        
        # 4. Make the prediction (will be a probability between 0 and 1)
        prediction_prob = model.predict(input_scaled)[0][0]
        
        # 5. Classify the prediction
        prediction = int(prediction_prob > 0.5) # 1 if > 0.5, else 0
        
        # 6. Return the result as JSON
        return jsonify({
            'potability': prediction,     # 0 or 1
            'probability': float(prediction_prob) # The raw probability
        })

    except Exception as e:
        return jsonify({'error': str(e)}), 400


if __name__ == '__main__':
    app.run(host="0.0.0.0", port=7860)