DataEngineer¶

from augini import DataEngineer
import pandas as pd

# Initialize with minimal configuration
engineer = DataEngineer(api_key="sk-...")

# Create sample data
df = pd.DataFrame({
    'CustomerID': ['C001', 'C002'],
    'Age': [25, 45],
    'MonthlyCharges': [50.0, 75.0]
})

# Generate a single feature
df = engineer.generate_feature(
    df=df,
    new_feature_name='customer_segment',
    new_feature_description='Classify customer into segments based on spending',
    output_type='category',
    constraints={'categories': ['Premium', 'Regular', 'Budget']}
)

# Generate occupation prediction
df = engineer.generate_feature(
    df=df,
    new_feature_name='PredictedOccupation',
    new_feature_description="Predict occupation based on age and spending",
    output_type='text'
)

# Example output:
#   CustomerID  Age  MonthlyCharges  PredictedOccupation
# 0      C001   25           50.0   Entry Level Professional
# 1      C002   45           75.0   Senior Manager

augmented_df = engineer.generate_features(
    df=df,
    features=[
        {
            'new_feature_name': 'ChurnRisk',
            'new_feature_description': 'Calculate churn risk score (0-100)',
            'output_type': 'float',
            'constraints': {'min': 0, 'max': 100}
        },
        {
            'new_feature_name': 'RetentionOffer',
            'new_feature_description': 'Suggest personalized retention offer',
            'output_type': 'text'
        }
    ],
    use_sync=False,  # Async processing for better performance
    show_progress=True
)

# Example output:
#   CustomerID  Age  MonthlyCharges  ChurnRisk  RetentionOffer
# 0      C001   25           50.0       35.5   "10% discount..."
# 1      C002   45           75.0       25.2   "Premium upgrade..."

engineer = DataEngineer(
    api_key="sk-...",
    model="gpt-4o-mini",  # Model selection
    temperature=0.8,      # Response creativity
    max_tokens=750,       # Max response length
    concurrency_limit=10, # Parallel processing limit
    base_url="https://openrouter.ai/api/v1",  # Custom API endpoint
    debug=False          # Debug mode
)

# Numeric output with constraints
df = engineer.generate_feature(
    df=df,
    new_feature_name='risk_score',
    new_feature_description='Calculate risk score',
    output_type='float',
    constraints={
        'min': 0,
        'max': 100,
        'step': 0.1
    }
)

# Categorical output with defined categories
df = engineer.generate_feature(
    df=df,
    new_feature_name='segment',
    new_feature_description='Customer segment',
    output_type='category',
    constraints={
        'categories': ['Premium', 'Regular', 'Budget']
    }
)

from augini import DataEngineer
import pandas as pd

# Initialize with configuration
config = {
    'api_key': 'your-api-key',
    'model': 'gpt-4-turbo-preview'
}

engineer = DataEngineer(config=config)

# Load your data
data = pd.read_csv('your_data.csv')

# Generate a new feature
result = engineer.generate_feature(
    df=data,
    new_feature_name='risk_score',
    new_feature_description='Calculate customer risk score based on transaction history',
    output_type='float'
)

# Generate a specific feature
feature_data = engineer.generate_feature(
    df=data,
    new_feature_name='customer_segment',
    new_feature_description="Create customer segments based on behavior",
    output_type='category',
    source_columns=['age', 'income', 'purchase_history']
)

# Generate multiple features at once
features_data = engineer.generate_features(
    df=data,
    new_feature_specs=[
        {
            'new_feature_name': 'lifetime_value',
            'new_feature_description': "Predict customer lifetime value",
            'output_type': 'float',
            'constraints': {'min': 0}
        },
        {
            'new_feature_name': 'churn_risk',
            'new_feature_description': "Assess customer churn risk",
            'output_type': 'category'
        }
    ]
)

# Generate feature with constraints
result = engineer.generate_feature(
    df=data,
    new_feature_name='satisfaction_score',
    new_feature_description='Calculate customer satisfaction score',
    output_type='float',
    constraints={
        'min': 0,
        'max': 100,
        'step': 0.1
    }
)

# Handle large datasets efficiently
result = engineer.generate_feature(
    df=large_data,
    new_feature_name='risk_score',
    new_feature_description='Calculate risk score',
    output_type='float',
    batch_size=32
)

config = {
    # Model settings
    'model': 'gpt-4o-mini',
    'temperature': 0.7,

    # Processing settings
    'batch_size': 32,
    'concurrency_limit': 5,

    # Performance settings
    'enable_cache': True,
    'cache_ttl': 3600
}