Pollsposition was created by Alex Andorra in 2017, and I joined in 2021 for the popularity dashboard. The idea is to understand (and predict) political trends using data and bayesian statistics. Models were originally implemented in PyMC, but I would like to re-implement them in Aesara. Aesara is more flexible, will hopefully have better (and faster) samplers for the kind of models we want to work with… and is in need of a (public) flagship project.

TODO Port model for president's popularity to Aesara

TODO Port the poll averaging model to Aesara

TODO Who do uncertain people vote for?

TODO Implement the model to compute pollster bias

TODO Ecological inference using election results

What is done in this draft for instance.


Links to this note