Do you have too many meaningless features?
Feature engineering is vital to AI success, especially for tabular data. Yet feature engineering is little more than a footnote in most popular machine learning education courses.
One of the challenges in teaching and practicing ML feature engineering is the lack of a systematic approach based on understanding of data semantics and database structure. As a result, feature lists are often extremely bloated, containing unexplainable features which are difficult to maintain and make sense of.
Discover a new framework for feature engineering and a set of signal types that intuitively explain their purpose, align with underlying data semantics, are mathematically rigorous, and inspire more creative feature engineering.
Explore Sergey Yurgenson’s slide deck from ODSC East 2023 below: