In this repo you can find end-to-end NLP project in which I downloaded reviews, analyzed them and created machine learning models for sentiment analysis.
Analysis and results of this project are described in docs https://github.com/kingagla/Reviews_classification/blob/main/docs/reviews%20classification.pdf
Please notice that to reproduce the project, folder roberta should contain pre-trained polish-roberta large model which is available here: https://github.com/sdadas/polish-roberta
Next you should run scripts with following order:
- Run all scripts from scripts/prepare_data folder with order suggested by names
- Run know_your_data_plots.py from scripts/create_plots
- Run scripts from scripts/models folder with order suggested by names