This repository contains all the necessary files and scripts for running, adapting, and testing machine learning models on financial datasets. Below is a description of the files and their respective purposes:
- Description: Contains the machine learning models.
- Purpose:
- Includes the code for building, training, and evaluating machine learning models on the Bloomberg dataset.
- Serves as the backbone for generating and fine-tuning models for financial market predictions.\
- Dependencies: Requires
FinancialMarketData.csvas the primary dataset.
- This folder contains all the pre-trained model
.pklfiles. - Each
.pklfile is a serialized machine learning model trained on the Bloomberg dataset for financial predictions. - Ensure this folder is not moved or renamed, as the scripts depend on its relative path.
- Description: The main code for the Streamlit app.
- Purpose:
- Serves as the user interface for portfolio allocation advice with ML Model
- Dependencies: Requires Streamlit and the
pkl_files/folder for loading models.
- Description: A script for testing the trained models on a synthetic dataset.
- Purpose:
- Adapts the trained models for use on synthetic datasets to evaluate their generalizability.
- Provides metrics and performance visualizations for backtesting.
- Dependencies: Requires the synthetic dataset and the models from the
pkl_files/folder.
- Description: The Bloomberg financial market dataset.
- Purpose:
- Acts as the training and testing dataset for the machine learning models.
- Ensure this file is in the root directory to avoid file path issues in
model.py.
- Place all the
.pklfiles inside thepkl_files/folder. - Run
run streamlit app.pyto start the Streamlit application and interact with the models.
pip install -r requirements.txt
For any questions or issues, please reach out to tpuvvala@gatech.edu