anomaly-grid-py 0.4.1

Deviation detection using variable-order Markov-chains in finite alphabet sequences

anomaly-grid-py

Deviation detection using variable-order Markov-chains in finite alphabet sequences.

Installation

pip install anomaly-grid-py

Usage

import anomaly_grid_py

# Create detector
detector = anomaly_grid_py.AnomalyDetector(max_order=3)

# Train on normal sequences
normal_sequences = [
    ['A', 'B', 'C'],
    ['A', 'B', 'D'],
    ['A', 'C', 'D']
] * 50  # Need enough data for training

detector.fit(normal_sequences)

# Get anomaly scores
test_sequences = [
    ['A', 'B', 'C'],  # Normal
    ['X', 'Y', 'Z']   # Anomalous
]

scores = detector.predict_proba(test_sequences)
print(scores)  # [0.26, 0.62] - higher = more anomalous

# Binary predictions
anomalies = detector.predict(test_sequences, threshold=0.5)
print(anomalies)  # [False, True]

API

# Create detector
AnomalyDetector(max_order=3)

# Train on normal sequences
detector.fit(sequences)

# Get anomaly scores [0,1]
detector.predict_proba(sequences)

# Get binary predictions
detector.predict(sequences, threshold=0.5)

# Get model info
detector.get_performance_metrics()

Requirements

• Python 3.8+
• NumPy

Development

git clone https://github.com/abimael10/anomaly-grid-py.git
cd anomaly-grid-py
./setup.sh
source venv/bin/activate
pytest tests/

License

MIT