INNER CODE UNIT · Python

create_sample_csv_data

balapriyac/data-science-tutorials · build-with-python/etl-pipeline/main.py:8

def create_sample_csv_data():
    """Create sample data and save as CSV to simulate downloading from a source"""
    np.random.seed(42)
    data = {
        'transaction_id': [f'TXN_{i:05d}' for i in range(1000)],
        'customer_id': np.random.randint(1, 201, 1000),
        'product_name': np.random.choice(['Laptop', 'Phone', 'Tablet', 'Headphones', 'Mouse'], 1000),
        'price': np.round(np.random.uniform(10, 1000, 1000), 2),
        'quantity': np.random.randint(1, 6, 1000),
        'transaction_date': pd.date_range('2024-01-01', periods=1000, freq='H'),
        'customer_email': [f'user{i}@email.com' if i % 10 != 0 else None for i in range(1000)]
    }
    df = pd.DataFrame(data)
    df.to_csv('raw_transactions.csv', index=False)
    print("Sample CSV file 'raw_transactions.csv' created!")
    return 'raw_transactions.csv'

def extract_data_from_csv(csv_file_path):

View source record →