This is the official working release, version 1.0, of the Stock Swing Predictor Tool.
I created this tool as a part of my award-winning Senior Thesis at Allegheny College. This release is essentially the same exact version of the tool that was presented as a part of the Thesis, with some minor improvements.
Feature List:
- Scrapes the amount of historical stock price data specified by the user
- Trains 7 models using scraped data, then makes price swing predictions
- Models: SVR-RBF, SVR-Linear, SVR-Polynomial, Linear Regression, Elastic Net, Lasso, K-Neighbors Regressor
- User can chose to use either a command-line or web app interface:
- Both interfaces graphically display how accurate models were in predicting the prices of historical days
- Both interfaces allow users to view the following results in a tabular format: future price swing predictions, future price predictions, model coefficient of determination scores, model efficiency times
- The web app was implemented in Streamlit and offers extra features when compared to the web UI, including the ability to get model predictions recommended based on the data size chosen
- Give user the option to export results data for future use
- Extensive Pytest test suite to ensure tool functionality
More improvements & official version releases will occur in the future.