Authors: Florian Eisenbarth, Nicolas Oulianov, Nicolas Boussenina, Armand Foucault, Eloi Alardet
This class group project is about predicting positions of deputies on votes in the French National Assembly.
We collect data from the government and heavily process it to build a custom supervized dataset.
We analyze this dataset to highlight political games inside the National Assembly.
Finally, using a Deep Learning model and smart feature engineering, we predict voting positions of a set of 10 political groups. We reach an accuracy of roughly 75%. We create a custom metric, a weighted F1 score tuned to our problem, that scores about 72% (best being 100%).
To get a better understanding of the data and the estimator, reading the starting kit notebook is a good start.
This repository is built to be a RAMP challenge. To learn about the RAMP library, please visit the documentation.
The submissions need to be located in the submissions folder. For instance for my_submission, it should be located in submissions/my_submission.
To run a specific submission, you can use the ramp-test command line:
ramp-test --submission my_submission
You can get more information regarding this command line:
ramp-test --help
This project is published under a MIT License.