A neural network that takes an input of a certain stock company and outputs a prediction about whether the stock value will increase or decrease based on news data about the company from that day.
inference.py is the file to be run in order to run the application, model.npy is the saved NN.
The RNN folder contains all neural network training files and scripts
The WebScraper folder contains all web scraping files and scripts
The TrainingData folder shows all the files collected in the training process using scraper.py
Uses the BeautifulSoup4 library for all webscraping
Uses the NewsApi library to locate top articles from the day containing a query
The webscraping and NewsAPI require the lxml library. This can be installed by rinning pip install lxml
Uses the MyGrad and MyNN libraries for the neural network forward pass and back propogation.
The Numba library is required to run inference. This can be installed by running pip install numba
Uses the Global Vector Library GloVe-50, which converts words to 50-dimensional vectors. The GloVe file can be downloaded here.
The Gensim library is required to load the GloVe library. This can be installed by running pip install gensim
The neural network achieved a final accuracy of approximately 64.3 percent