posteriordb: a database with data, models and posteriors

Models and data play a crucial role in the development and improvements of posterior inference algorithms and statistical research. We present posteriordb, a database of posteriors with models and data. Currently, the database contains roughly 100 posteriors with a wide variety of different properties, such as multimodality, high-dimensionality, and funnel-shaped posterior geometries. We aim for the posteriordb to become a Stan community project for sharing models and data in a robust and well-documented fashion, both using Stan code as well as other probabilistic programming frameworks. The use cases of the database range from the development of new posterior approximation techniques, teaching, empirical experimentation, statistical research, to unit testing of different probabilistic programming frameworks.

Documentation: https://github.com/MansMeg/posteriordb