OpenML-AutoML Tutorial JADS

Schedule

Time	Topic
13:15 - 14:45	Introduction to AutoML
15:00 - 16:30	AutoML Hands-on

There are several options:

JADS Jupyter environment
- Download (or clone) this github repository. Locate the notebook folder.
- In your JADS Jupyter environment, create a new folder, click the upload button, and upload everything from the notebook folder into that new folder
- Open and run the Python setup notebook to install all dependencies
- Verify that everything installs correctly
Binder
- Click the Binder button on the website to start a new environment on myBinder. This will start a docker image, so it may take a few minutes.
Google Colab
- We provided links to Google Colab notebooks on the website. You need to make your own copy of each of them via File > Save Copy in Drive
Set up your own environment
- OpenML part: Instructions are available in the beginning of the OpenML Python tutorial
- AutoML part: Do the setup as in the Python setup notebook. Note: Autosklearn does NOT run on Windows

Create an OpenML account on http://www.openml.org - this is needed for uploading things to the platform
If you have interesting data to work with, please bring it along.
- Open datasets can be shared during the hands-on session.
- Any canonical machine learning problems (classification/regression) can be experimented with during the AutoML hands-on session.