Use of Bert Algorithm to predict the sentiment of financial tweets

NLP-ENSAE

This repository contains the project of Bertrand VUILLEMOT and Nathan BRY for the class Machine Learning for NLP at ENSAE.

We conduct an analysis and trained a BERT model on a dataset containing about 28k tweets on financial topics (stocks, cryptos, etc). The Data can be found here.

• The report describing the different steps of the project can be found in NLP_ENSAE_Bry_Vuillemot.pdf
• The notebook containing the code (analysis + models) can be found in NLP_project.ipynb