Overview
Tilmann Gneiting is a statistician known for foundational work on probabilistic forecasting and its evaluation. At the time of Gneiting–Raftery (2007) he was Associate Professor of Statistics at the University of Washington, Seattle; he later led the Computational Statistics group at the Heidelberg Institute for Theoretical Studies (HITS) and held a chair at the Karlsruhe Institute of Technology. His research established the modern framework of proper scoring rules, calibration and sharpness, and probabilistic weather forecasting.
Key Contributions / Features
- Gneiting-Raftery (2007): With Adrian E. Raftery, the definitive review and development of proper scoring rules — the convex-function/Bregman-divergence characterization, the CRPS and its multivariate generalization the energy score, the interval score, and the connection of the logarithmic score to Bayes factors and to optimum-score (M-)estimation. See Gneiting-Raftery (2007).
- "Sharpness subject to calibration": articulated (with Balabdaoui and Raftery, and with Westveld) the paradigm that the goal of probabilistic forecasting is to maximize the sharpness of the predictive distribution subject to its calibration against outcomes.
- CRPS and the energy score: popularized the continuous ranked probability score as a standard, honesty-enforcing metric for continuous and multivariate probabilistic forecasts.
- Spatial statistics and covariance functions: known also for work on nonseparable space–time covariance functions and Matérn-type models (outside the scope of the sources currently in this wiki).
Related