Model Monitor

A comprehensive Python library for automated model monitoring and drift detection in machine learning systems.

Features

• Data Drift Detection: Identify shifts in feature distributions using statistical tests and distance metrics
• Model Performance Monitoring: Track key metrics over time and detect degradation
• Prediction Drift Analysis: Monitor changes in model output distributions
• Customizable Alerting: Set thresholds and receive notifications when drift is detected
• Visualization Tools: Generate comprehensive reports and visualizations of drift metrics
• Integration Flexibility: Works with various ML frameworks and data sources
• Scalable Architecture: Efficiently handle large datasets and high-throughput models

Installation

pip install model-monitor

Quick Start

from model_monitor import Monitor
import pandas as pd

# Initialize monitor with baseline data
baseline_data = pd.read_csv("baseline_data.csv")
production_data = pd.read_csv("production_data.csv") 

monitor = Monitor()
monitor.set_baseline(baseline_data)

# Run drift detection
results = monitor.detect_drift(production_data)

# Generate report
monitor.generate_report("drift_report.html")

# Set up automated monitoring
monitor.schedule(
    data_source="s3://bucket/path/to/data/",
    frequency="daily",
    alert_threshold=0.05,
    notification_email="alerts@example.com"
)

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

Author

• Biswanath Roul - GitHub

License

This project is licensed under the MIT License - see the LICENSE file for details.