outdatedLabs 0.1.0

A secure package for training machine learning models

OutDatedLabs - Secure Machine Learning Training Package

A Python package for secure machine learning model training using the ML training server.

Installation

pip install outdatedLabs

Quick Start

from outdatedLabs import SecureModel

# Create a linear regression model
model = SecureModel.linearRegression()

# Optionally set a custom server URL
model.set_server_url("http://your-server:8000")

# Train the model
model.fit(
    dataset_hash="your_dataset_hash",
    features=["feature1", "feature2"],
    target="target_column"
)

# Make predictions
predictions = model.predict(X)

# Get training metrics
metrics = model.get_metrics()
print(metrics)

Features

• Secure model training using encrypted datasets
• Support for linear regression models
• Progress tracking with tqdm
• Comprehensive error handling and logging
• Easy-to-use interface
• Configurable server URL

API Reference

SecureModel

The main class for secure model training.

Methods

• linearRegression(server_url: str = "http://localhost:8000") -> SecureModel

  • Create a new linear regression model instance

• set_server_url(server_url: str) -> SecureModel

  • Set a custom URL for the ML training server

• fit(dataset_hash: str, features: List[str] = None, target: str = None, params: Dict[str, Any] = None) -> SecureModel

  • Train the model using the specified dataset

• predict(X: Union[pd.DataFrame, List[List[float]]]) -> List[float]

  • Make predictions using the trained model

• score(X: Union[pd.DataFrame, List[List[float]]], y: List[float]) -> Dict[str, float]

  • Calculate model performance metrics

• get_metrics() -> Dict[str, Any]

  • Get training metrics

Requirements

• Python 3.7+
• requests
• joblib
• pandas
• scikit-learn
• tqdm

License

MIT License