generate_model.py

import tensorflow as tf
import numpy as np
import os

# ==========================================
# Digital Will - Inactivity Model Generator
# ==========================================
# This script generates a custom TFLite model for the Digital Will App.
# Input: 18 features (User activity metrics)
# Output: 3 classes (Active, Suspicious, Deceased)

def generate_model():
    print("🚀 Generating Synthetic Data...")
    
    # generate 1000 samples
    samples = 1000
    
    # Feature inputs (randomized for synthetic training)
    # 0-4: Days since activity (Post, Like, Comment, DM, AppOpen)
    # 5-9: Averages (Posts30, Posts90, Likes30, Likes90, PostVariance)
    # 10-14: Interaction (ActiveHours, FollowerRate, DmResponse, Sessions, SessionDuration)
    # 15-17: consistency (Battery, Location, Time)
    
    X = np.random.rand(samples, 18).astype(np.float32)
    y = np.zeros((samples, 3)).astype(np.float32)

    for i in range(samples):
        # Logic to determine labels based on "days since app open" (index 4)
        days_inactive = X[i][4] * 200 # scale to 0-200 days
        
        if days_inactive < 30:
            y[i] = [1.0, 0.0, 0.0] # Active
        elif days_inactive < 90:
            y[i] = [0.2, 0.8, 0.0] # Suspicious
        else:
            y[i] = [0.0, 0.1, 0.9] # Deceased/Inactive

    print("🧠 Building Model...")
    model = tf.keras.Sequential([
        tf.keras.layers.Dense(32, activation='relu', input_shape=(18,)),
        tf.keras.layers.Dropout(0.2),
        tf.keras.layers.Dense(16, activation='relu'),
        tf.keras.layers.Dense(3, activation='softmax') # 3 classes
    ])

    model.compile(optimizer='adam', loss='categorical_crossentropy', metrics=['accuracy'])
    
    print("🏋️ Training Model...")
    model.fit(X, y, epochs=10, batch_size=32, verbose=0)

    print("💾 Converting to TFLite...")
    converter = tf.lite.TFLiteConverter.from_keras_model(model)
    tflite_model = converter.convert()

    # Ensure directory exists
    output_dir = "assets/ml_models"
    if not os.path.exists(output_dir):
        os.makedirs(output_dir)

    output_path = f"{output_dir}/inactivity_model.tflite"
    with open(output_path, "wb") as f:
        f.write(tflite_model)
    
    print(f"✅ SUCCESS! Model saved to: {output_path}")
    print("You can now run the app with full AI features.")

if __name__ == "__main__":
    try:
        generate_model()
    except Exception as e:
        print(f"❌ Error: {e}")
        print("Please ensure tensorflow is installed: pip install tensorflow")