cattleia: Complex Accessible Transparent Tool for Learning Ensemblers In AutoML

cattleia is a Dash application for analysing model ensembles produced by AutoML packages in Python: auto-sklearn, AutoGluon and FLAML.

🌸 through tables and visualizations, allows to look at the metrics of the models to assess their contribution to the prediction of the built committee

🌸 introduces compatimetrics, which enable analysis of model similarity

🌸 enables the analysis of weights and how they impact the model by allowing their modification

Application Setup Guide

System Requirements

• Linux or WSL for Windows
• Python >= 3.8

Instruction

1. Clone the repository:

   git clone https://github.com/malwina0/cattleia.git

2. Install dependencies:

   cd cattleia
   pip3 install -r requirements.txt

3. Start the app:

   python3 app.py

4. Access the Dash app at http://127.0.0.1:8050/. Keep the server running to use the app.

Motivation

The motivation for the research is to understand the inner workings of ensemble models and to develop new insights and techniques that can be used to improve the performance of AutoML models in tabular data prediction. The cattleia aims to improve the trustworthiness and interpretability of AutoML. The goal is to make this tool intuitive to use, the same time providing valuable information and analysis.

App demo

Authors

This project is created as a Bachelor thesis by:

• Dominik Kędzierski,
• Jakub Piwko,
• Malwina Wojewoda.

Project co-ordinator and supervisor: Anna Kozak.