An introduction to Scikits-Learn

This is the September CamPUG meeting. Normally some of us go on to the pub afterwards.

Corran Webster from Enthought will talk about Scikits-Learn.

Scikits-Learn is an open-source library for machine learning in Python. In this talk you will learn the basic concepts of Scikits-Learn: how to prepare your data, how to run classification and regression on your data sets, how to validate your results, as well as some common pitfalls to avoid. As part of the talk, you'll build a classifier of handwritten digits. By the end of the talk you will have an overview of the way that Scikits-Learn's API works and be able to use it to create your own machine learning models.

People will need to bring laptops (although we should be working in groups). Corran says:

For the packages that will be needed, I think we're good with reasonably recent versions of:

• scikits-learn
• pandas
• matplotlib
• jupyter notebooks

and their dependencies (ie. numpy, scipy).

Using Enthought's deployment manager (which is freely available) from http://docs.enthought.com/edm/ you can create a self-contained environment for the session, which won't mess up an existing Python environment, and will have all needed dependencies, with:

edm install -e campug scikits_learn pandas matplotlib jupyter

and then enter that environment with either:

edm activate -e campug

or in a self-contained shell with:

edm shell -e campug

Or alternatively, folks can install using pip or other Python environment tools.

Enthought have kindly said they will be providing pizza for this talk.