LIDA: Science of Data Science
The Science for Data Science Programme is LIDA’s newest programme. We aim to build an interdisciplinary network of researchers with a common interest in the underpinning science and practice of data science, covering issues such as open science, reproducibility, causal inference, data equity, and algorithmic interpretability. We welcome anyone with an interest in the fundamental theory and methods of data science.
Background and aims
Data science is a diverse and rapidly evolving area. Each year, new ideas, techniques, and principles emerge that change the scope and norms of the field. Keeping up-to-date with these developments requires a broad and reflective focus on the changing science and practice of data science itself. The Science for Data Science seeks to respond to this evolving landscape by spotlighting and exploring the theory, methods, and practice of data science. In doing so, we aim to support more robust, transparent, and impactful research across the University of Leeds.
Working with other networks and communities across Leeds, our main ambition is to host enlightening and thought-provoking events, focussed on exploring new knowledge, ideas, and developments in the field.
Recent activities
The Science of Data Science programme launched in June with an energetic event on “Meeting the AI and big data revolutions in quantitative social science”. Featuring provocative talks from Dr Peter Tennant (School of Medicine), Prof Dan Birks (School of Law), and Prof Mandeep Dhami (Middlesex University), the event debated the risks and benefits of using current and near-future AI tools in quantitative social science research. Dr Tennant opened by warning the audience of the limits of current AI technologies, raising concerns around bias and transportability. Prof Birks and Prof Dhami offered robust examples of where AI has helped to examine new types of social science data on hitherto unmanageable scales.
Future events
In our next event, on 1st December 2025, Dr Rachel Hughes from University of Bristol will be introducing quantitative bias analysis (QBA), a form of sensitivity analysis focussed on using external data to better understand and quantify the impact of residual biases. Although the roots of QBA stretch back many decades, the approach is rapidly gaining popularity due to new methods and the open science revolution. Further details to follow.
Community news
Co-director Roger Beecham has published a new book on Visualization for Social Data Science described by Andrew Gelman (Department of Statistics, Columbia University) as “an important book on an important topic”!
Find out more about LIDA: Science of Data Science Return to the Annual Showcase 2025