Predictions

A project to recognize and assess prediction data (text, numerical, audio, visual). The goal is to provide an analysis of how accurate predictions are.

Table of Contents

• Definitions
• File Structure
• Installation
• Setup NaviGator
• Usage
• Contributing

File Structure

├── misc                        # Contains random pieces of unfinished code.
├── prediction_classification   # Contains the pipeline to classify if a sentence is a prediction or not. 
├── prediction_correctness      # Contains the pipeline to assess how similar a prediction is to an actual outcome.
├── classification_models.py    # Contains the models to classify if a sentence is a prediction or not. 
├── clean_predictions.py        # Contains the code to clean our data.
├── data_processing.py          # Contains the code to manipulate our data.
├── feature_extraction.py       # Contains the code to extract features from predictions.
├── log_files.py                # Contains the code to produce a log file.
├── requirements.py             # Contains the requiremmts to run code in project.
├── text_generation_models.py   # Contains the LLMs to generate our data.
└── README.md                   # Project documentation

Installation

Use the package manager you prefer. If uv package manager, follow the below

• Fork the repo and see latest work in development branch, unless stated otherwise by one of the contributors.

• Use the package manager you prefer. If uv package manager, follow the below.

1. Install the uv package manager. For macOS, you can use brew install uv,
2. OPTIONAL: Create a project with uv init . that'll default to name of directory. It may need to be repository name predictions, so you could try uv init predictions
   • If you already see a .toml file, you should be able to skip.
3. Create virtual environment with uv venv or uv venv <name> (uv venv .venv_predictions)
4. Activate virtual environment with source .venv/bin/activate or source .<name>/bin/activate (source .venv_predictions/bin/activate)
5. Install requirements with uv pip install -r pyproject.toml
6. Install uv pip install ipykernel so you can run the jupyter notebooks
7. Create a .env file
   • Create a NaviGator Toolkit API key -- NAVI_GATOR_API_KEY = "djb2". See steps below.
   • OPTIONAL: Create a Groq Cloud API key -- GROQ_CLOUD_API_KEY = "djb". Similar to NaviGator steps.

Setup NaviGator

Only UF students

1. Navigate to NaviGator Toolkit.
2. Enter your UF log in credentials.
3. Click Virtual Keys $\rightarrow$ + Create New Key.
   1. For Team, select "navigator-toolkit", which should be the default option.
   2. Enter a Key Name of your choice. An example is uf_data_studio_predictions_project or predictions_project.
   3. For Models, you can select All Team Models
   4. You can enter/skip Optional Settings.
   5. Click Create Key
4. Ensure you have forked the repo, then navigate to the development branch, create a new file called '.env', and add the string NAVI_GATOR_API_KEY= <your_api_key>.

Usage

See the issue you're working on for details. If no details are provided, then reach out (dj.brinkley@ufl.edu unless we already have another mode of communication).

 - Create a [NaviGator](https://api.ai.it.ufl.edu/) API key -- `NAVI_GATOR_API_KEY = "djb2"`
 - OPTIONAL: Create a [Groq Cloud](https://console.groq.com/) API key -- `GROQ_CLOUD_API_KEY = "djb"`