Auto ML

The tryout

Hello and welcome back! Today we’ll see an example of AutoML and what it can do. I will be using Auto-Sklearn as an example and might explore others in the future.

Today we will start with installing Auto-Sklearn and I will explain some errors I came across and how I troubleshoot them. I am using Jupyter Notebook for all of this as well as pip and homebrew.

Firstly I started by installing auto-sklearn by using pip.

$ pip install auto-sklearn

This should work according to the auto-sklearn documentation. Unfortunately, that was not the case for me. When I tried running it I got the installation error:

Apparently, wheel is an installation package that is needed for the installation. When I tried using pip install wheel it did not work. I decided to switch to homebrew to complete the installation.

$ brew install wheel

After installing wheel I ran into a different problem when trying to install auto-sklearn.

Therefore swig needed to be installed.

$ brew install swig

And after all of that, auto-sklearn finally was installed. But when it needed to run, surprise surprise an error occurred on the Jupyter Notebook.

To fix this issue, I decided to uninstall and reinstall scikit-learn to be able to remove this error, and presto! all done no more errors.

Now, just as a fair warning updating scikit-learn may mess up packages that are already installed and that you may currently use. In my case, the package that is now unable to be used is my lazypredict package.

There you go! You are now ready to use auto-scikit learn!