RoboBuffett
RoboBuffett copied to clipboard
Machine Learning project to predict stock performance based on textual analysis. Uses Bayesian multinomial classifier.
Project goal: Predict future stock performance based on textual analysis of SEC filings.
Modules:
Scraper (Izaak): Pull SEC documents from EDGAR.
Stock (Ahmad): Get stock prices and maintain CIK<->Ticker mapping.
Manager (Dan): Manage & organize data. Manages parsing, serialization, classification, and testing.
Parser (Dan): Parse SEC filings, getting header info and document text.
Classifier (Ahmad): Generate training classifications for documents, calls on Stock. Called by manager.
To be added: Multinomial model generator, LLV (log likelihood value) classifier. To be maintained by Dan.