Machine Learning for Time Series

About

Time series data is essential in fields like finance, energy, healthcare, and climate science. This course covers time series forecasting and anomaly detection, focusing on sequential patterns in univariate and multivariate data. Participants will learn to choose appropriate models, apply best practices, and adapt machine learning methods for accurate predictions, uncertainty estimation, and anomaly detection in diverse time series challenges.

Course leader

Dr Y. Yuliya Shapovalova

Target group

Advanced Bachelor, Master, PHD, Postdoc, Professional.

Admission requirements
- Basic knowledge or willingness to catch up with the basics of probability theory (in particular, familiarity with concepts like Gaussian/normal distribution)
- Basic knowledge of mathematics and statistics (concepts like mean, variance, probability distribution)
- Understanding of basic modelling approaches such as regression/classification
- Basic knowledge of Python is necessary for practical tasks (e.g., familiarity with libraries like numpy, pandas, scipy, matpliolib).

Course aim

1. Understand Time Series Fundamentals
2. Understand Anomaly Detection Problem
3. Train and Evaluate Machine Learning Models
4. Handle Uncertainty Using Probabilistic Methods
5. Preprocess Data and Do Feature Engineering
6. Work with Real Data and Avoid Common Pitfalls

Fee info