Project: MTSAnomalyDetection (GitHub Link)

MTS Anomaly Detection

Multidimensional Time Series Anomaly Detection (MTSAD)

Glossary

Anomaly (+1) -> Anomalies -> Anomaly Detection
Normal (-1)
Observed
Predicted

Data

MTS² Format

t	[features]	tag
t₀	[v₀, v₁, ..., v_d-1]	-1 / 'n'
t₁	[v₀, v₁, ..., v_d-1]	+1 / 'a'
...	...	...

Columns:

t: datetime.
[features]: d features.
tag: +1 or a for anomaly, -1 or n for normal.

MTS² Format : Multidimensional Time Series Supervised (MTSS)

Note: All datasets should be transformed into this format for further processing.

Algorithm

Supervised

Semi-Supervised

Unsupervised

System Architecture

License

MTSAnomalyDetection is licensed under the Apache License, Version 2.0. See LICENSE for the full license text.