@grafana/scenes-ml

Add Machine Learning functionality to your Scenes.

What is this?

This library contains a collection of @grafana/scenes objects which can be added to your Scenes to run interactive, responsive machine learning algorithms directly in the browser.

@grafana/scenes-ml currently contains implementations of the following:

• forecasting (using the MSTL/ETS algorithms)
• outlier detection (using the median absolute difference or DBSCAN algorithms)
• changepoint detection (using either Bayesian Online Changepoint Detection or Autoregressive Gaussian Process Changepoint Detection)

Under the hood, the heavy lifting is largely powered by the augurs library, which runs inside WebAssembly. See that library for more information on the underlying algorithms.

Usage

See the library README for usage documentation.

Development

To work on @grafana/scenes-ml, please follow the guides below.

Setting up local version of @grafana/scenes-ml with an app plugin

1. Run YARN_IGNORE_PATH=1 yarn link from packages/scenes-ml directory.
2. Run yarn dev from packages/scenes-ml directory.
3. Run yarn link @grafana/scenes-ml from app plugin directory.
4. Start app plugin development server.

Demo app

Alternatively, use the demo app included in this repository.