Fueling innovative software
July 15-18, 2019
Portland, OR

K-means for anomaly detection

Anais Dotis (InfluxData)
4:15pm4:55pm Thursday, July 18, 2019
Secondary topics:  AI Enhanced
Average rating: ****.
(4.67, 3 ratings)

Who is this presentation for?

  • Developers, data scientists, and data analysts




Everyone seems eager to incorporate ML in their anomaly detection solutions, but it doesn’t always make sense to use ML. Using statistical methods to detect one-off peaks in time series data is extremely effective and efficient. However, statistical methods fail when trying to detect contextual or collective anomalies.

If you want to alert on these anomalies, you probably need to use some type of ML. Anais Dotis-Georgiou outlines three statistical methods, introduces two popular ML algorithms, and describes the relationship between contextual and collective anomalies. Join in to learn how easy it is to use k-means to alert on a contextual anomaly in EKG data.

Prerequisite knowledge

  • A working knowledge of Python

What you'll learn

  • Learn three popular statistical methods for anomaly detection and two types of time series anomalies where those methods will fail
  • Understand the four different ML categories
  • See how to use k-means to alert on EKG data with scikit and InfluxData Python CLI
Photo of Anais Dotis

Anais Dotis


Anais Dotis-Georgiou is a developer advocate at InfluxData with a passion for making data beautiful using data analytics, AI, and machine learning. She takes the data that she collects and does a mix of research, exploration, and engineering to translate the data into something of function, value, and beauty. When she’s not behind a screen, you can find her outside drawing, stretching, or chasing after a soccer ball.