- encoding - Contains the data aggregation and pre-processing for Reddit posts and the NLP (Doc2Vec) model that trains the Reddit post data. Training the model is quite lengthy (30+ minutes) so we avoid running it unless you wish to build a new model (a previous model file is in the Google Drive). Fetching the Reddit data of the 10 stock tickers is also quite lengthy as well.
- rnn - Contains the recurring neural network (LSTM) model. This is the file that should be run. It requires three pickle files and the price CSV for each stock ticker: the vectorized posts, daily upvotes, daily volume, and the CSV containing the closing price. This will output confusion matrices of the training and test set's accuracy.
- stock_api - Fetches the financial data from the Polygon API from a start and end date. Only used for data aggregation.
https://drive.google.com/drive/folders/1Myg18mgqhjtY-EXWZ9nO2uqRkXT_lhAC