Financial-Sentiment-Analysis

Setup

0. Create a file newsapi.key and paste your news API key in the first line.

Option A - conda

1. Install miniconda from https://docs.conda.io/en/latest/miniconda.html
2. Install dependencies conda env create -f environment.yml for conda or conda create --name finbert --file requirements.txt
3. Download model from link and put it in default models directory finbert/models/sentiment/base/
4. Install nltk dependencies

import nltk
nltk.download('punkt')

Option B - docker

1. docker build -t finbert .
2. docker run -it --rm finbert:latest -p 8050:8050 -p 8051:8051

Run

1. conda activate finbert
2. python3 main.py Input date if manual else it'll fetch data for today.
3. python3 app.py for Dashboard.

Note : The terminal will output the url for each dashboard

Screenshots

Credits

Couldn't have done this without the ProsusAI/finBERT repo ofcourse.

@misc{1908.10063,
Author = {Dogu Araci},
Title = {FinBERT: Financial Sentiment Analysis with Pre-trained Language Models},
Year = {2019},
Eprint = {arXiv:1908.10063},
}