Zero to Machine Learning in 3 Steps

Machine Learning“Machine learning is 20 percent fun, 80 percent elbow grease.”

This is what one client told Dinesh Nirmal, vice president of analytics development for IBM Analytics. There’s no doubt about the fun part. Machine learning promises to revolutionize many aspects of IT. It’s that other 80 percent you don’t hear much about.

At its recent machine learning launch, IBM took some of the mystery away. There are many interviews worth checking out, but one in particular caught our attention. In it, Dinesh takes machine learning from concept to application through three pillars. The entire interview is worth watching to get all the details, but here’s a summary of the three pillars to get you started:

  1. Product

As Dinesh says, “You have to have a product, right?” This is the very basis of getting started with machine learning, getting that product with different shared sets of functions and features that let customers build their models of what they want to do.

  1. Process

Of course, once you have your model, you actually have to make it happen. “There’s the process of taking that model that you built in a notebook and being able to operationalize it,” is how Dinesh summarizes it.

This is not easy! Vendors and service providers try to make it sound easy, but there is a lot involved. Let’s look at the example of a large enterprise. You’ve built your model; now you have to deploy it into your infrastructure or production environment. If you’re like many large companies, this infrastructure has likely been evolving for decades. You might have third-party software you can’t change. You might even have legacy applications that were written more than 30 years ago. (Raise your hand if this is you.) Nobody wants to mess with this stuff. This brings us to…

  1. People

This pillar could also be called expertise, experience, skills or any of the other things a company needs to complete the process of making machine learning work. If yours is like most companies, you probably don’t have a ton of expertise in executing machine learning strategies.

The ultimate takeaway here is: get help. Make use of the resources available to you and get a partner as you go down this path. Our partnership with IBM is a good place to start.

Ready to make machine learning work for you? Give us a call.

Drew Woods
Key Information Systems, Inc.