custom-pyod

High-performance outlier detection algorithms using Polars.

Features

• COPOD: Copula-Based Outlier Detection
• ECOD: Empirical Cumulative Distribution Based Outlier Detection
• Built on Polars for high performance
• Compatible with scikit-learn API

Installation

pip install custom-pyod

Quick Start

from custom_pyod.models import COPOD, ECOD
import polars as pl

# Load your data as a Polars DataFrame
df = pl.DataFrame({
    'feature1': [1, 2, 3, 100],
    'feature2': [1, 1, 1, 50]
})

# COPOD
copod = COPOD()
outlier_scores = copod.fit_predict(df)

# ECOD
ecod = ECOD()
outlier_scores = ecod.fit_predict(df)

License

MIT License