In this video we will walk though the Netdata Anomaly Advisor deepdive python notebook.

The aim of this notebook is to explain, in detail, how the unsupervised anomaly detection in the Netdata agent actually works under the hood. No buzzwords, no magic, no mystery :)

Notebook: https://github.com/netdata/netdata/blob/master/ml/notebooks/netdata_anomaly_detection_deepdive.ipynb

Try it for yourself, get started by signing in to Netdata and connecting a node. Once initial models have been trained (usually after the agent has about one hour of data, zero configuration needed), you'll be able to start exploring in the Anomaly Advisor tab of Netdata.

We'd love any feedback as you try this new feature out. Please feel free to leave feedback in the Netdata community, discord, GitHub discussions or just drop us an email at analytics-ml-team@netdata.cloud.

• https://community.netdata.cloud/
• https://github.com/netdata/netdata/discussions
• https://discord.gg/mPZ6WZKKG2

00:00 - Introduction

01:15 - Running notebook in Colab

02:15 - Deepdive overview

03:30 - Preprocessing

06:57 - Getting some raw data

08:31 - Add some anomalous data

09:41 - ML discussion

18:07 - Visualizing the anomaly score

20:24 - How it actually works

20:40 - Heatmap visualization

23:45 - Lineplot visualisation

24:15 - Barplot visualisation

25:02 - Scatterplot visualisation

26:05 - Wrapping up