data_generation.py

import pandas as pd
import numpy as np
import random
from datetime import datetime, timedelta

def generate_synthetic_data(num_customers=1000, random_seed=42):
    """
    Simulates realistic customer financial data with behavioral patterns leading to financial stress.
    """
    np.random.seed(random_seed)
    random.seed(random_seed)
    
    data = []
    
    for i in range(1, num_customers + 1):
        customer_id = f"CUST_{i:04d}"
        
        # Base attributes
        monthly_salary = np.random.choice([25000, 35000, 50000, 75000, 100000, 150000]) * np.random.uniform(0.9, 1.1)
        loan_emi_amount = monthly_salary * np.random.uniform(0.1, 0.4) # EMI is 10-40% of salary
        
        # Determine if this customer will miss a payment (approx 45% financial stress rate for demonstration)
        missed_payment = 1 if np.random.random() < 0.45 else 0
        
        if missed_payment:
            # Behavioral patterns for financially stressed customers
            salary_credit_delay_days = np.random.randint(3, 15) # Salary is delayed
            weekly_bank_balance = monthly_salary * np.random.uniform(0.01, 0.1) # Low balance
            upi_tx_count = np.random.randint(30, 80) # High number of small transactions
            upi_tx_amount = monthly_salary * np.random.uniform(0.6, 0.9)
            
            spending_food = monthly_salary * np.random.uniform(0.3, 0.5)
            spending_entertainment = monthly_salary * np.random.uniform(0.01, 0.05) # Cut down on entertainment
            spending_bills = monthly_salary * np.random.uniform(0.1, 0.2)
            
            utility_payment_timing = np.random.choice(["Late", "On-Time"], p=[0.7, 0.3])
            atm_withdrawals_count = np.random.randint(5, 12) # Increased ATM withdrawals
            atm_withdrawals_amount = monthly_salary * np.random.uniform(0.15, 0.3)
            
            upi_lending_apps_flag = 1 if np.random.random() < 0.6 else 0 # High chance of using lending apps
            
            # Trend features over last 4 weeks (simulated)
            balance_change_wow_pct = np.random.uniform(-0.5, -0.1) # Balance dropping 10-50% week over week
            spending_drop_ratio = np.random.uniform(0.3, 0.8) # Significant drop in overall spending
            
        else:
            # Behavioral patterns for healthy customers
            salary_credit_delay_days = np.random.randint(-1, 2) # Salary on time or 1 day late/early
            weekly_bank_balance = monthly_salary * np.random.uniform(0.3, 1.5) # Healthy balance
            upi_tx_count = np.random.randint(15, 50)
            upi_tx_amount = monthly_salary * np.random.uniform(0.3, 0.6)
            
            spending_food = monthly_salary * np.random.uniform(0.15, 0.3)
            spending_entertainment = monthly_salary * np.random.uniform(0.1, 0.25)
            spending_bills = monthly_salary * np.random.uniform(0.1, 0.2)
            
            utility_payment_timing = np.random.choice(["Early", "On-Time", "Late"], p=[0.4, 0.5, 0.1])
            atm_withdrawals_count = np.random.randint(0, 4)
            atm_withdrawals_amount = monthly_salary * np.random.uniform(0.01, 0.05)
            
            upi_lending_apps_flag = 1 if np.random.random() < 0.1 else 0 # Low chance
            
            balance_change_wow_pct = np.random.uniform(-0.05, 0.1) # Stable or growing balance
            spending_drop_ratio = np.random.uniform(0.9, 1.1) # Consistent spending
            
        data.append({
            "customer_id": customer_id,
            "monthly_salary": round(monthly_salary, 2),
            "salary_credit_delay_days": salary_credit_delay_days,
            "weekly_bank_balance": round(weekly_bank_balance, 2),
            "upi_tx_count": upi_tx_count,
            "upi_tx_amount": round(upi_tx_amount, 2),
            "spending_food": round(spending_food, 2),
            "spending_entertainment": round(spending_entertainment, 2),
            "spending_bills": round(spending_bills, 2),
            "utility_payment_timing": utility_payment_timing,
            "atm_withdrawals_count": atm_withdrawals_count,
            "atm_withdrawals_amount": round(atm_withdrawals_amount, 2),
            "loan_emi_amount": round(loan_emi_amount, 2),
            "upi_lending_apps_flag": upi_lending_apps_flag,
            "balance_change_wow_pct": round(balance_change_wow_pct, 4),
            "spending_drop_ratio": round(spending_drop_ratio, 4),
            "missed_payment": missed_payment
        })
        
    df = pd.DataFrame(data)
    return df

if __name__ == "__main__":
    df = generate_synthetic_data()
    print(df.head())
    print(f"Total shape: {df.shape}")
    print(f"Financial Stress Rate: {df['missed_payment'].mean():.2%}")