https://github.com/aws-samples/simple-anomaly-detection-solution

https://github.com/aws-samples/simple-anomaly-detection-solution

Last synced: about 2 months ago
JSON representation

• Host: GitHub
• URL: https://github.com/aws-samples/simple-anomaly-detection-solution
• Owner: aws-samples
• License: mit-0
• Created: 2021-08-25T03:46:46.000Z (about 3 years ago)
• Default Branch: main
• Last Pushed: 2024-01-12T18:21:13.000Z (9 months ago)
• Last Synced: 2024-05-30T02:45:49.794Z (4 months ago)
• Language: Python
• Size: 241 KB
• Stars: 6
• Watchers: 8
• Forks: 2
• Open Issues: 1
• Metadata Files:
  • Readme: README.md
  • Contributing: CONTRIBUTING.md
  • License: LICENSE
  • Code of conduct: CODE_OF_CONDUCT.md

Awesome Lists containing this project

README

        
# Simple Anomaly Detection Solution

This is an anomaly detection solution that helps user to quickly identify anomaly in time series data.

In a typical time series analysis use case, this solution provides an easy way to quick analyze a subset of data with anomaly to quickly evaluation a couple of models during exploratory data analysis. This give user a sense of time series data quality and whether there is anomaly pattern in the dataset.

A simple user interface built on top of Streamlit to provide a quick glance on the analysis result.

## Supported Models

- Univariate Inter Quantile Ratio

- Multivariate Vector Auto Regression

- Multivariate Isolation Forest

- Multivariate Mahalanobis Distance

## Usage

This application has been tested with python 3.9.

```bash

git clone https://github.com/yapweiyih/simple-anomaly-detection

cd simple-anomaly-detection

# Activate virtual environment first and install python package, this may take a while, so take a cup of coffee.

pip install -e .

pip install -r requirements.txt

# streamlit run src/uc_timeseries/streamlit_app.py -- --data_dir 

streamlit run src/uc_timeseries/streamlit_app.py -- --data_dir refdata

```

## Demo

A demo data with some anomaly points has been included to get you familarized with this tool.

### Model Selection

Select desired model to run training and evaluation.

![Model selection](images/model_selection.png)

### Evaluation

Visualize time index with anomaly deteceted and feature importance for tree based model.

![Evaluation](images/evaluation.png)

## Security

See [CONTRIBUTING](CONTRIBUTING.md#security-issue-notifications) for more information.

## License

This library is licensed under the MIT-0 License. See the LICENSE file.