Anomaly Detection in Time Series Using Python

May 28, 2020 / Xebia Academy Webinar Week

Vadim Nelidov - 28 May 9:30 - 10:30 AM (CET)


From inventory to website visitors, resource planning to finance and accounting, much of any organization’s data has temporal nature. Key problems facing data scientists dealing with time series include not only forecasting the future values, but also identifying when these values are alarming. Intertemporal anomaly detection often becomes a complex problem that neither common rule-based heuristics nor standard anomaly detection algorithms can tackle well. In this webinar we will have a close look at this domain, learn about its distinctive features and challenges as well as learn to detect anomalies in time series data, while dodging many common mistakes and issues.

Why attend this webinar?

If you want to broaden your understanding of time series data, extra challenges that it provides and learn to tackle them when it comes to anomaly detection – this webinar is for you. If you already have experience in the field, you may still want to reconsider some of the practices that you use and pick up some new ones. If you deal with other domains of Data Science, this may still be an interesting showcase of how complex data-driven problems can be solved in Python.


Key takeaways

  • How anomalies in time series and their detection are different from time-invariant data
  • Why rule-based heuristics may be a poor solution
  • Retrospective vs forward-looking approaches
  • Why some models suitable for regular forecasting might not be suitable for anomaly detection
  • How to tell a good anomaly detection system from a poor one
  • How to build robust detection systems in Python

Who is it for?

This webinar is best suited for people having some background in Data Science or Statistics. Interest and affinity with time series and forecasting is a plus. Some intermediate knowledge of Python would help benefit the most from the webinar, though it can still be useful for those experienced in statistical frameworks other than Python

Speaker: Vadim Nelidov


Vadim is an enthusiastic Data Science specialist fluent in deciphering data-driven problems. He is passionate about sharing his knowledge and insights, believing that Data literacy should not be a privilege of a few. And his goal is to be there to make this a reality. Making the intricacies of Data analysis intelligible and uncovering the regularities hiding in the data is a major source of his inspiration.

Vadim has a long-lasting relationship with Data Science and Analytics. Originally from Moscow, he has his professional roots in mathematics, statistics and economics. Since his Bachelor years, he has already been involved in analytical competitions and tutoring. In 2014 he moved to Netherlands for his Masters and PhD, which he is now happy to call his home. Over these years, Vadim gained invaluable experience working with diverse data sources and methods, doing innovative research in multinational teams, and training and supervising dozens of students. Sharing knowledge has become a true passion for him as well as finding the right approach that works for everyone.

In his work, Vadim sees far beyond what is on the surface and gets to the essence of the problems, discovering long term data-driven solutions rather than temporary fixes. Having an academic background, he is always at the frontier of the latest techniques and innovations. This allows him to understand,  optimise and interpret cutting-edge algorithms that may be merely a black box for others.

For the quote: “One of my favourite moments is to see a sparkle in people’s eyes when they face surprising insights that you uncovered in their data. Knowledge that was always there, but remained hidden until a proper key was found.”

Sign up now