https://github.com/HPI-Information-Systems/DADS

Distributed detection of sequential anomalies in univariate time series
https://github.com/HPI-Information-Systems/DADS

Last synced: about 2 months ago
JSON representation

Distributed detection of sequential anomalies in univariate time series

• Host: GitHub
• URL: https://github.com/HPI-Information-Systems/DADS
• Owner: HPI-Information-Systems
• Created: 2020-07-20T13:21:12.000Z (about 4 years ago)
• Default Branch: master
• Last Pushed: 2021-01-05T10:43:43.000Z (over 3 years ago)
• Last Synced: 2024-06-01T14:33:17.761Z (4 months ago)
• Language: Java
• Homepage:
• Size: 838 KB
• Stars: 9
• Watchers: 8
• Forks: 2
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# Distributed Detection of Sequential Anomalies in Univariate Time Series

Source code for the Distributed Anomaly Detection System (DADS)

This algorithm is based on the [Series2Graph](http://helios.mi.parisdescartes.fr/~themisp/series2graph/) algorithm, that was published in `P. Boniol and T. Palpanas, Series2Graph: Graph-based Subsequence Anomaly Detection in Time Series, PVLDB (2020)`.

## Usage

### Building

Requirements:

- JDK 8 (Java)

- Maven

```sh

mvn package

```

You can skip the tests with `-DskipTests`.

The created fat JAR is located in the `target`-folder.

### Notes

The algorithm only accepts binary input files without timestamps and ordered double values.

To convert a text file, use the Python script in [`script/sequence-converter/convert.py`](./script/sequence-converter/convert.py).

#### Example Conversion

Input file:

```text

0.333443

0.466437

0.474765

```

Converting:

```shell

python3 script/sequence-converter/convert.py --input input.txt --output output.bin

```

### Running

Requirements:

- JRE 8 (Java)

```sh

java -jar .jar master --host localhost --port 7788 --min-slaves 0 \

    --sequence  \

    --sub-sequence-length 50 --intersection-segments 50 \

    --query-length 75 --convolution-size 16 \

    --output ./results.txt --no-statistics

```