automl sandbox

demo.mov

About

Created during a 2-day Hackathon mainly for the purpose of learning more about currently available Auto ML techniques. It also serves as a proof of concept of how easily Auto ML libraries like autosklearn and autokeras (etc.) can be exposed through a web interface, something which can benefit everyone, from data scientists and analysts, to those with less knowledge about machine learning but with more domain knowledge.

Install backend

# auto-sklearn failed for me on python=3.10 so ensure you have python=3.9
conda create -n "automl39" python=3.9
conda activate automl39

Option A. Install automatically from our requirements.txt

pip install -r requirements.txt

Or, if that doesn't work, use:

pip install -r requirements_all.txt

Option B. Install "manually"

# Install auto-sklearn dependencies from a remote requirements file
curl https://raw.githubusercontent.com/automl/auto-sklearn/master/requirements.txt | xargs -n 1 -L 1 pip install

# Finally install auto-sklearn
pip install auto-sklearn

Install client (for interacting with the backend)

Assuming you have node and npm installed:

cd ./src/web
npm i

Run it

Backend:

# Terminal tab 1
cd ./src/web
npm run start:backend

Client:

# Terminal tab 1
cd ./src/web
npm run start:client