Skip to content

Machine learning model for predicting the likelihood of success for short activist reports targeting public companies.

License

Notifications You must be signed in to change notification settings

DanGlChris/short_activist_predictor

Repository files navigation

Short activist predictor

Note

This Python package provides a machine learning model for predicting the likelihood of success for short activist reports targeting public companies.

Installation

You can install the package using pip:

pip install short_activist_predictor

Usage

The package provides a Predictor module with function predict_pdf() that takes in a path to the PDF file of the short report and the target company name. It loads and analyzes the report text using NLP. The Predictor then outputs probability scores between 0-1 predicting the chance of a successful outcome from the report release over 3 time periods - within 1 days, 1 week, and 1 month.

Here's an example of how to use the Predictor.predict_pdf() function:

from short_activist_predictor.predictor import Predictor

# Create a predictor instance by hf token indentification
predictor_ = Predictor(Predictor.Login_Token())

# Upload the report in pdf format
# This following function will ask you to upload a pdf to
while(True):
  predictor_.predict_pdf()

To request access, please email [email protected] with your name, institution affiliation, and details on your proposed use case. We will evaluate requests and provide access to those with legitimate needs aligned with the intended uses of this model

Requirements

To use this package, you need to have Python 3.10 or higher installed on your system. You also need to have the following packages installed:

  • Bertopic
  • Transformers
  • Huggingface_hub
  • NLTK

License

This package is licensed under the GNU General Public License v3.0. See the LICENSE file for more information.

Contact

If you have any questions or suggestions, feel free to contact Daglox Kankwanda at @danglchris.

About

Machine learning model for predicting the likelihood of success for short activist reports targeting public companies.

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages