PyKAN WebUI

If you like this project, please give it a ⭐️ on GitHub!

A modern, modular Streamlit web platform for symbolic regression using PyKAN.

Features

• Upload CSV dataset
• Select independent/dependent variables and configure PyKAN parameters
• Start training with progress and logs
• Visualize results and compare test set predictions
• Modular multi-page design for maintainability
• English UI and code comments
• Engineering best practices: docstrings, modularity, session utils, config, and tests

Directory Structure

.
├── app.py                # Home page and global config
├── pages/                # Each page is a separate module
│   ├── 1_upload_data.py
│   ├── 2_configure_parameters.py
│   ├── 3_train_model.py
│   └── 4_visualization.py
├── src/                  # Core logic modules
│   ├── data_utils.py
│   ├── config.py
│   ├── pykan_runner.py
│   ├── kan_dictionary.py
│   ├── feature_engineering.py
│   └── plot_utils.py
├── utils/                # Utility functions (e.g., session)
│   └── session.py
├── tests/                # Unit tests
├── requirements.txt      # Dependencies
├── .env                  # Environment/config (optional)
└── README.md

Usage

pip install -r requirements.txt
streamlit run app.py --server.port 8080 --server.address 0.0.0.0

Demo

You can try the live demo here: https://pykanwebui.streamlit.app/

Extending

• Add new pages in pages/