.. image:: https://img.shields.io/pypi/v/kenchi.svg :target: https://pypi.org/project/kenchi

.. image:: https://img.shields.io/pypi/pyversions/kenchi.svg :target: https://pypi.org/project/kenchi

.. image:: https://img.shields.io/pypi/l/kenchi.svg :target: https://github.com/HazureChi/kenchi/blob/master/LICENSE

.. image:: https://img.shields.io/conda/v/Y_oHr_N/kenchi.svg :target: https://anaconda.org/Y_oHr_N/kenchi

.. image:: https://img.shields.io/conda/pn/Y_oHr_N/kenchi.svg :target: https://anaconda.org/Y_oHr_N/kenchi

.. image:: https://img.shields.io/readthedocs/kenchi/stable.svg :target: http://kenchi.rtfd.io

.. image:: https://img.shields.io/travis/HazureChi/kenchi/master.svg :target: https://travis-ci.org/HazureChi/kenchi

.. image:: https://img.shields.io/appveyor/ci/Y-oHr-N/kenchi/master.svg :target: https://ci.appveyor.com/project/Y-oHr-N/kenchi/branch/master

.. image:: https://img.shields.io/coveralls/github/HazureChi/kenchi/master.svg :target: https://coveralls.io/github/HazureChi/kenchi?branch=master

.. image:: https://img.shields.io/codeclimate/maintainability/HazureChi/kenchi.svg :target: https://codeclimate.com/github/HazureChi/kenchi

.. image:: https://mybinder.org/badge.svg :target: https://mybinder.org/v2/gh/HazureChi/kenchi/master?urlpath=lab

kenchi

This is a scikit-learn compatible library for anomaly detection.

Dependencies

• Required dependencies #. numpy>=1.13.3 <http://www.numpy.org/>_ (BSD 3-Clause License) #. scikit-learn>=0.20.0 <http://scikit-learn.org/>_ (BSD 3-Clause License) #. scipy>=0.19.1 <https://www.scipy.org/scipylib/>_ (BSD 3-Clause License)
• Optional dependencies #. matplotlib>=2.1.2 <https://matplotlib.org/>_ (PSF-based License) #. networkx>=2.2 <https://networkx.github.io/>_ (BSD 3-Clause License)

Installation

You can install via pip

::

pip install kenchi

or conda.

::

conda install -c y_ohr_n kenchi

Algorithms

• Outlier detection #. FastABOD [#kriegel08]_ #. LOF [#breunig00]_ (scikit-learn wrapper) #. KNN [#angiulli02], [#ramaswamy00] #. OneTimeSampling [#sugiyama13]_ #. HBOS [#goldstein12]_
• Novelty detection #. OCSVM [#scholkopf01]_ (scikit-learn wrapper) #. MiniBatchKMeans #. IForest [#liu08]_ (scikit-learn wrapper) #. PCA #. GMM (scikit-learn wrapper) #. KDE [#parzen62]_ (scikit-learn wrapper) #. SparseStructureLearning [#ide09]_

Examples

.. code:: python

import matplotlib.pyplot as plt
import numpy as np
from kenchi.datasets import load_pima
from kenchi.outlier_detection import *
from kenchi.pipeline import make_pipeline
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import StandardScaler

np.random.seed(0)

scaler = StandardScaler()

detectors = [
    FastABOD(novelty=True, n_jobs=-1), OCSVM(),
    MiniBatchKMeans(), LOF(novelty=True, n_jobs=-1),
    KNN(novelty=True, n_jobs=-1), IForest(n_jobs=-1),
    PCA(), KDE()
]

# Load the Pima Indians diabetes dataset.
X, y = load_pima(return_X_y=True)
X_train, X_test, _, y_test = train_test_split(X, y)

# Get the current Axes instance
ax = plt.gca()

for det in detectors:
    # Fit the model according to the given training data
    pipeline = make_pipeline(scaler, det).fit(X_train)

    # Plot the Receiver Operating Characteristic (ROC) curve
    pipeline.plot_roc_curve(X_test, y_test, ax=ax)

# Display the figure
plt.show()

.. figure:: https://raw.githubusercontent.com/HazureChi/kenchi/master/docs/images/readme.png :align: center

References

.. [#angiulli02] Angiulli, F., and Pizzuti, C., "Fast outlier detection in high dimensional spaces," <https://doi.org/10.1007/3-540-45681-3_2>_ In Proceedings of PKDD, pp. 15-27, 2002.

.. [#breunig00] Breunig, M. M., Kriegel, H.-P., Ng, R. T., and Sander, J., "LOF: identifying density-based local outliers," <https://doi.org/10.1145/335191.335388>_ In Proceedings of SIGMOD, pp. 93-104, 2000.

.. [#dua17] Dua, D., and Karra Taniskidou, E., "UCI Machine Learning Repository," <https://archive.ics.uci.edu/ml>_ 2017.

.. [#goix16] Goix, N., "How to evaluate the quality of unsupervised anomaly detection algorithms?" <https://arxiv.org/abs/1607.01152>_ In ICML Anomaly Detection Workshop, 2016.

.. [#goldstein12] Goldstein, M., and Dengel, A., "Histogram-based outlier score (HBOS): A fast unsupervised anomaly detection algorithm," KI: Poster and Demo Track, pp. 59-63, 2012.

.. [#ide09] Ide, T., Lozano, C., Abe, N., and Liu, Y., "Proximity-based anomaly detection using sparse structure learning," <https://doi.org/10.1137/1.9781611972795.9>_ In Proceedings of SDM, pp. 97-108, 2009.

.. [#kriegel11] Kriegel, H.-P., Kroger, P., Schubert, E., and Zimek, A., "Interpreting and unifying outlier scores," <https://doi.org/10.1137/1.9781611972818.2>_ In Proceedings of SDM, pp. 13-24, 2011.

.. [#kriegel08] Kriegel, H.-P., Schubert, M., and Zimek, A., "Angle-based outlier detection in high-dimensional data," <https://doi.org/10.1145/1401890.1401946>_ In Proceedings of SIGKDD, pp. 444-452, 2008.

.. [#lee03] Lee, W. S, and Liu, B., "Learning with positive and unlabeled examples using weighted Logistic Regression," In Proceedings of ICML, pp. 448-455, 2003.

.. [#liu08] Liu, F. T., Ting, K. M., and Zhou, Z.-H., "Isolation forest," <https://doi.org/10.1109/ICDM.2008.17>_ In Proceedings of ICDM, pp. 413-422, 2008.

.. [#parzen62] Parzen, E., "On estimation of a probability density function and mode," <https://doi.org/10.1214/aoms/1177704472>_ Ann. Math. Statist., 33(3), pp. 1065-1076, 1962.

.. [#ramaswamy00] Ramaswamy, S., Rastogi, R., and Shim, K., "Efficient algorithms for mining outliers from large data sets," <https://doi.org/10.1145/335191.335437>_ In Proceedings of SIGMOD, pp. 427-438, 2000.

.. [#scholkopf01] Scholkopf, B., Platt, J. C., Shawe-Taylor, J. C., Smola, A. J., and Williamson, R. C., "Estimating the Support of a High-Dimensional Distribution," <https://doi.org/10.1162/089976601750264965>_ Neural Computation, 13(7), pp. 1443-1471, 2001.

.. [#sugiyama13] Sugiyama, M., and Borgwardt, K., "Rapid distance-based outlier detection via sampling," <http://papers.nips.cc/paper/5127-rapid-distance-based-outlier-detection-via-sampling>_ Advances in NIPS, pp. 467-475, 2013.