pyanom

This library is Python projects for anomaly detection. This contains these techniques.

Kullback-Leibler desity estimation
Singular spectrum analysis
Graphical lasso
CUMSUM anomaly detection
Hoteling T2
Directional data anomaly detection

REQUIREMENTS

numpy
pandas
scikit-learn
scipy

INSTALLATION

pip install pyanom

USAGE

Kullback-Leibler desity estimation

import numpy as np
from pyanom.density_ratio_estimation import KLDensityRatioEstimator

X_normal = np.loadtxt("./data/normal_data.csv", delimiter=",")
X_error = np.loadtxt("./data/error_data.csv", delimiter=",")

model = KLDensityRatioEstimator(
   band_width=h, lr=0.001, max_iter=100000)
model.fit(X_normal, X_error)
anomaly_score = model.score(X_normal, X_error)

Singular spectrum analysis

import numpy as np
from pyanom.subspace_methods import SSA

y_error = np.loadtxt("./data/timeseries_error2.csv", delimiter=",")

model = SSA(window_size=50, trajectory_n=25, trajectory_pattern=3, test_n=25, test_pattern=2, lag=25)
model.fit(y_error)
anomaly_score = model.score()

Graphical lasso

import numpy as np
from pyanom.structure_learning import GraphicalLasso

X_normal = np.loadtxt("./data/normal_data.csv", delimiter=",")
X_error = np.loadtxt("./data/error_data.csv", delimiter=",")

model = GraphicalLasso(rho=0.1)
model.fit(X_normal)
anomaly_score = model.score(X_error)

Direct learning sparse changes

from pyanom.structure_learning import DirectLearningSparseChanges

model = DirectLearningSparseChanges(
   lambda1=0.1, lambda2=0.3, max_iter=10000)
model.fit(X_normal, X_error)
pmatrix_diff = model.get_sparse_changes()

CUSUM anomaly detection

import numpy as np
from pyanom.outlier_detection import CAD

y_normal = np.loadtxt(
   "./data/timeseries_normal.csv", delimiter=",").reshape(-1, 1)
y_error = np.loadtxt(
   "./data/timeseries_error.csv", delimiter=",").reshape(-1, 1)

model = CAD(threshold=1.0)
model.fit(y_normal)
anomaly_score = model.score(y_error)

Hoteling T2

import numpy as np
from pyanom.outlier_detection import HotelingT2

X_normal = np.loadtxt("./data/normal_data.csv", delimiter=",")
X_error = np.loadtxt("./data/error_data.csv", delimiter=",")

model = HotelingT2()
model.fit(X_normal)
anomaly_score = model.score(X_error)

Directional data anomaly DirectionalDataAnomalyDetection

import numpy as np
from pyanom.outlier_detection import AD3

X_normal = np.loadtxt(
   "./data/normal_direction_data.csv", delimiter=",")
X_error = np.loadtxt("./data/error_direction_data.csv", delimiter=",")

model = AD3()
model.fit(X_normal, normalize=True)
anomaly_score = model.score(X_error)

Development Status
- 4 - Beta
License
- OSI Approved :: MIT License
Operating System
- OS Independent
Programming Language

Release files for pyanom 0.0.2b1

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for pyanom 0.0.2b1
File	Size	Uploaded
pyanom-0.0.2b1.tar.gz	10.4 kB	Sep 7, 2021	Details

Built distribution (wheel)

Table of built distributions (wheels) for pyanom 0.0.2b1
File	Interpreter	ABI	Platform	Reset
pyanom-0.0.2b1-py3-none-any.whl 12.4 kB Sep 7, 2021	Python 3	none	any	Details

Total release size: 22.7 kB

Release files / pyanom-0.0.2b1.tar.gz

Download URL	pyanom-0.0.2b1.tar.gz
Size	10.4 kB
Tags	Source
SHA-256 checksum How to use checksums	`f4c476040f1a72f83742933bd49db2c833ac58f88f48e76f3db97729bdf2546c`
BLAKE2b-256 checksum How to use checksums	`a5f8e422d15cb2812b94125523472dc40a9aa00564722bcebe8b7ac4042c9c02`
Upload date	Sep 7, 2021
Uploaded using Trusted Publishing? What is trusted publishing?	No
Uploaded via	`twine/3.4.2 importlib_metadata/4.8.1 pkginfo/1.7.1 requests/2.25.1 requests-toolbelt/0.9.1 tqdm/4.59.0 CPython/3.7.10`

Release files / pyanom-0.0.2b1-py3-none-any.whl

Download URL	pyanom-0.0.2b1-py3-none-any.whl
Size	12.4 kB
Tags	Python 3
SHA-256 checksum How to use checksums	`0dc0801fe0528e4c4d306e33692a96eea9ff070a44fc2ffc6ec00200004c8298`
BLAKE2b-256 checksum How to use checksums	`fffb7f145a626a433d66d97784b7a07e0e24678346810ede26852549a34ba5e8`
Upload date	Sep 7, 2021
Uploaded using Trusted Publishing? What is trusted publishing?	No
Uploaded via	`twine/3.4.2 importlib_metadata/4.8.1 pkginfo/1.7.1 requests/2.25.1 requests-toolbelt/0.9.1 tqdm/4.59.0 CPython/3.7.10`