MLNE - curriculum and format

[Thomas Merritt-Smith]

Hi everyone

Following yesterday's meetup, which was again really well attended, I wanted to get some feedback on where people want the group to go next, in terms of what we should cover and how we should cover it. I felt that the session yesterday was too talk heavy (my fault - I didn't really have a good dojo problem prepared), and am keen to keep future sessions as practical as possible. That said, there is a balance to be struck between 'shove the data into the black box' hyper-practical Machine Learning, and 'let me dust off my textbook' hyper-theoretical learning. 

At the moment, my plan is to do a bit of general 'up-skilling' in some of the key ML algorithms and techniques (regression, some of the classifiers, cross-validation and loss functions) in the hope that then we'll be able to move on as a group into more active stuff, mini-competitions on datasets, etc. etc. 

I think we could roughly follow the 'Introduction to Statistical Learning' table of contents

Please leave any comments here. I'm interested in the following: 

- Anyone interested in doing a talk

- What your level of interest/expertise is

- Whether you think the 'upskilling' plan is reasonable, or whether we should just jump in at the deep end

- Do you love/hate 'Introduction to Statistical Learning'?

I think it might be nice to have a full-on dojo on Linear Regression since we didn't do much actual coding yesterday. Let me know if you agree!

Once again, thanks to Dom and Geektalent for the Datascience talk and the pizza, and of course Campus North for being awesome. And beer.

Cheers

TMS