Advanced Amazon Redshift: Analytics and Amazon Machine LearningGo to Lab
I'd like to some info about understanding how the modeling is actually working and explainability.
it was all just following steps but nothing about *why* we were doing anything we were doing
could speed up lab by doing things in parallel, or give us more background to read about the problem set and what ML is really doing while we're waiting for the many minutes of time for the system to finish creating various tasks
Should not be using Windows (lame). Also, no-one cares about setting up the JDBC driver for SQL Workbench. This step should already be completed so there is more time to do the lab. Also, there should be more of "what is the point" behind this lab instead of scrambling to paste commands into SQL Workbench to complete the lab in time, without really understanding what you're accomplishing at the high level. Basically, this was a waste of time, I could've done all of this online on my own without paying for this overcrowded conference. This is the worst conference I've ever been too. You'd think it would be a better experience given the cost, and the fact that my company pays AWS over 100k per month
It was very easy to follow. It makes Machine Learning seem crazy easy.
It's a bit unclear as to why we used a Redshift cluster here over a more traditional DBMS, but other than that it was interesting. I do wish more details on what the ML configuration was doing.
It would be nice to have someone explaining why you do certain things or the pros/cons of one way versus another. The steps were very "monkey see, monkey do" and didn't have much of this type of explanation in the instructions. ALSO, the final SQL statement should have JOINed back to the flight data so that we could see which dates/hours/etc were the most/least likely to have delays.
Good lab, need larger machines or smaller data sets, lots of waiting