Learn directly from Valeriy Manokhin, PhD, MBA, CQF

As a researcher focusing on the modeling and simulation of technical systems, I frequently develop ML pipelines for high-stakes environments - such as sensor fusion for security systems and environmental forecasting. In these domains, a model must be reliable and work in the real world. I enrolled in this course to systematically integrate Conformal Prediction into my research, and Dr. Valeriy Manokhin’s curriculum exceeded all expectations. The course is densely packed with valuable information, covering both the historical development and modern applications of Conformal Prediction across all major ML fields. Valeriy’s teaching style is excellent; he goes far beyond just reading dense slides. He perfectly balances rigorous scientific formulas with easy-to-understand examples, frequently pausing to share his own invaluable expertise and honest opinions on different approaches. Having the ability to discuss complex, edge-case questions directly with a leading expert in the field added immense value to the learning experience. Since I am already utilizing his books in my work, it was a distinct privilege to take this class. There is a unique pedagogical weight to learning this material from a researcher in the direct academic lineage of Vladimir Vovk and Andrey Kolmogorov. Highly recommended for any researcher or data scientist who needs to deploy trustworthy machine learning models in the real world.

I'm grateful to Valery for the comprehensive overview of Conformal Prediction methods and their applications. The course provided a solid introduction to the framework and its main ideas. Personally, I would have appreciated more discussion around practical implementation details and real-world use cases, especially for specific applied questions.

Deep and well-structured coarse. Provides material above the standard level, gives an integral view of the time series topic + lots of comments and examples based on the vast experience in the industry.

The course was good! I enjoyed learning about modern forecasting techniques. However, I believe the course may be too advanced for people with no prior forecasting experience. You need to have at least some foundational knowledge of forecasting to follow along effectively; otherwise, it’s easy to get lost quickly.

One of the most practically grounded forecasting courses I've come across. The progression from forecastability and classical models (ARIMA, ETS) through ML evaluation, EDA, deep learning, and conformal prediction feels deliberate and well-paced, each session builds on the previous one rather than standing alone. Standout moments: the anecdote about rebuilding a forecasting system across 70 countries with only 104 data points is worth the price of admission on its own. The DLinear paper deconstruction, showing a simple linear MLP outperforming complex transformer architectures, is a healthy reality check on current hype. The conformal prediction session covers genuinely cutting-edge material that you rarely find in applied courses. Overall, a course that takes the practitioner seriously: no hand-waving on evaluation, no uncritical enthusiasm for the latest architectures. Strongly recommended for anyone moving beyond tutorial-level forecasting.