Editorials

Machine Learning, or I have so much to learn!

Have you worked with any machine-learning scenarios? With SQL Server? There are so many options and opportunities for starting work with SQL Server in particular and with interesting data flows in general.

It’s a bit overwhelming the types of things you can do – from working across database platforms to analyze information to machine learning right in the SQL Server platform, to mixtures of these, where different types of systems do different types of things.

A few times I’ve written about whether big data is paying off, and the premise that it’s (not yet) producing. I don’t buy it, but I think that comes down to not taking the time to ask questions of the information or use tools that can help point out those interesting questions and topics.

I can honestly say that, without exception, when I’ve taken the time to ingest various workloads and make them available to any number of processes, from content analysis to predictive stuff to trying to see patterns in things that have happened historically, I’ve not ever been disappointed. I might find out things are NOT even related, or that one thing doesn’t necessarily lead to another (often in spite of what I’ve been spouting off about at the time), but I have always come away with at least a tidbit, and often an outright lightbulb type of learning.

Boy is the learning curve steep. If you’re working with SQL Server, databases in general and getting up to speed on the types of things you can do, how they’re done, the options for analysis and all of that, the ways you have of going about that, both from a learning and an execution standpoint, are many. It’s hard to choose a direction. We’ve said many times that SQL Server in general is very broad and few, if any, know all of the moving pieces. Rather most choose to specialize and make a go of it that way.

The same is true of learning R or working with multiple platforms with common data flows. Pick a direction and run like crazy.

What types of things have you seen done, or are you doing, with the machine learning? I was reading about some of the things rolling around out there with SQL Server and such here in this post from TechTarget and I’m very curious what types of things people are doing that…

… they would consider “machine learning”
… they’ve had success with
… they’ve had some challenges with

What have you seen?