This is a notebook and some supporting files for analysing local election results following on from the first Campaign Campaign Lab event.
The idea is to predict the results of local elections at ward-level using some of the new data that people have been collating, which should yield some intereisting insights into what factors can be used to predict the results and where local campaigning has been particularly effective.
You should be able to run the notebook as long as you have installed reasonably up to date versions of the following python libraries:
- jupyter (I used version 1.0.0) for running the interactive notebook
- pandas (0.23.1) for doing neat operations with tables of data
- pystan (2.17.1.0) a python interface to Stan, a statistical modelling platform
- matplotlib (2.0.0)for drawing graphs
You’ll also need to download some ONS data about income levels in local authorities from here and save it as `data/income_data.csv`.