Skip to content

training LSTM models to predict stock index prices from sequential historical data and financial news sentiment

Notifications You must be signed in to change notification settings

2021sshah/BigDataFinalProject

Repository files navigation

Big Data Final Project

Training LSTM models to predict stock index prices from sequential historical data and financial news sentiment

This project utilizes LSTMs and GRUs with both news sentiment and historical time-series stock index movement as features to predict the daily closing price of stock indexes (Starting with DJIA).

The ProsusAI FinBERT pre-trained transformer model was used to extract sentiment from each article in the following financial news dataset. These were aggregated and averaged to obtain sentiment scores for each day, which were included in the predictive model. The cleaned and extracted daily sentiment data can be found in the Financial News Sentiment CSV.

About

training LSTM models to predict stock index prices from sequential historical data and financial news sentiment

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Contributors 3

  •  
  •  
  •