AI Model Training Release
This release introduces the Stochastic Gradient Descent classifier, an AI model we are utilizing to make buy, sell, or hold predictions. In particular, this release contains new code that is used to train the model based on stock trend reports with a buy / sell / hold label determined by finite metrics. This can be seen in the ai_train.py file, included in the auto_label_entry function.
Work Accomplished:
- Optimized Selenium code to reduce the number of times the program has to connect / disconnect to the webdriver.
- Added and improved logging to let the user know how the program is operating.
- Wrote the ai_train.py file, which processes the trend report, labels it with a buy / sell / hold label, and feeds it to the AI model to train it.
- Added the news headlines section. The code reads headlines regarding a certain stock and feeds it to a financial sentiment AI (Finbert) to get a positive / negative / neutral outlook.
- Bug fixes and connection issue fixes.
To-do list:
- Train the model extensively with trend reports from different stocks daily (this can also be completed by the user in the future if they desire).
- [Adding from last release] Start the AI analysis/prediction implementation.
- [Adding from last release] Consider how the reflect option will work.
Screenshots of discussed code:
