NLP Analyse of election machine texts from the Finnish parliament elections 2015.
The data used for the prediction comes from avoindata.fi: https://www.avoindata.fi/data/fi/dataset/eduskuntavaalien-2015-ylen-vaalikoneen-vastaukset-ja-ehdokkaiden-taustatiedot
The result of the analysis is that the model could make party predictions of any given candidates "I want to work towards"- statemanet with approx. 44% accuracy on the test data. A complete report and statistics can be found under reports: https://github.com/chpatola/election_nlp/tree/master/reports
- Clone this repository to your own computer
- cd to the repository
- Create a new conda environment where is the name you want to give the new environment. Then activate the enivronment
conda create --name <env> --file requirement.txt
conda activate <env>
-
Set your base path at line 7 in predict_model.py
-
Run the code from the file predict_model.py.
python predict_model.py
- The programme will print some output of the analysis but you will also find results in the reports folder: https://github.com/chpatola/election_nlp/tree/master/reports