Ivan Shubin
Implementing anomaly detection for time series can be challenging, with many techniques and tools available. But can you achieve effective results without AI or Machine Learning? In this talk, we will demonstrate how basic statistical methods can effectively detect anomalies in time series data. We'll show you how to use Grafana to visualize these anomalies on graphs and ensure past incidents do not impact future predictions. Additionally, we will explore building Grafana dashboards as code as part of the anomaly detection solution and adjusting the detection for various events.
Ivan Shubin[node:field-speakers-institution]
Hi, my name is Ivan. I am a Senior Site Reliability Engineer at Booking.com. Before that I worked at TomTom and eBay. Throughout my career, I have explored various roles including Quality Assurance, Software Engineering, System Administration, and SRE. I have always been fascinated by the complexity of high-load and distributed systems and have a passion for understanding how everything works. In my spare time, I enjoy working on my open-source project, Schemio, which I use to build various interactive visualisations on SRE topics.
author = {Ivan Shubin},
title = {Anomaly Detection in Time Series from Scratch Using Statistical Analysis},
year = {2024},
address = {Dublin},
publisher = {USENIX Association},
month = oct
}