Learn by doing: how to think about and apply machine learning

The 3 job categories topping the list of increasing demand in the 2020 World Economic Forum Future of Jobs Report involve data, machine learning, and artificial intelligence. Here are some anecdotal observations about machine learning to illustrate how data analytics/machine learning/AI are transforming many other jobs.

I've partnered with leading universities around the country to teach and advise students on making an impact at work. In the past 3 years, I've seen the use of machine learning in student capstone projects go from exotic (machine learning as a project focus) in a very small subset of projects (<10%) to routine (machine learning as a tool in a typical project) in a steadily increasing fraction of projects (over 20% and climbing fast). I see more and more students reach for machine learning to solve their problems almost as naturally as they graph data. I expect very soon, machine learning will become another set of tools in the toolkit of any quantitatively competent professional. (Digitally fluent professional? 

Just as today we expect these professionals—in social sciences, business and finance, physical sciences, computer science, engineering, applied math—to be able to plot data, curve fit, use descriptive statistics, and straightforward statistical tests, we will increasingly expect all professionals that use data to make decisions or guide their work to be comfortable with machine learning tools. 

Supporting the premise: machine learning is popping up as canned routines that are GUI-accessible in Matlab, Minitab, and other programs. Do you remember exporting data from spreadsheets to graphing programs? ML add-ins to spreadsheets are already here and we will see ML tools in spreadsheet programs themselves very soon.

Here is one great way to boost your machine learning know-how. I highly recommend Andrew Ng's "Introduction to Machine Learning."

I completed this course earlier this year, and would give it the subtitle "Learn by Doing: How to Think About and Apply Machine Learning." I highly recommend it to anyone from practitioners to technical or senior leadership.

https://www.coursera.org/learn/machine-learning/home/welcome

This course hits the mark with a wide audience. The more you bring to the course the more you'll be able to get from it, while at the same time it is approachable, so you can get from it what you need.

You can take it without a linear algebra background, and especially if you've done the simplest coding you'll be able to work your way through. If you have a background in linear algebra you'll find it easy sailing in most parts so your focus can be thinking about machine learning. Have experience with numerical methods, especially optimization? Get ready for fun. If you've thought about validating models and experimental design, you will really enjoy the parts about training, cross-checking, and validating models (analyzing the incremental value of more data, more features, sensitivity to ML model search parameters, ...). Towards the end of the course, you'll learn how to organize and lead solo work or teams developing complex machine learning workflows. 

Pro-tip: take advantage of the free license to use online Matlab for this course. The project notebooks make completing the work much easier, and as Andrew Ng points out, Matlab is a common way to prototype ML algorithms.

I know many people have started their machine learning journey with Kaggle. That is a great option. I think starting with the course, and moving to Kaggle, as I have, is a great way to go. There are probably as many pathways available as there are people -- it is a personal journey in many respects.

I'd enjoy hearing your thoughts about who needs ML skills and how to get them. Thanks!

Bessel combines human centered design and agile teamwork to transform individuals, teams, and organizations. 

About Chris Danek

Chris has spent the past 25 years building successful medical device companies. He created Bessel Origin based on years of teaching and research in partnership with teams at Santa Clara University, Catholic University in Washington, D.C., and The University of Texas at El Paso. Chris is an engineer, entrepreneur, and teacher. He studied design thinking and new product and process development while earning his PhD from Stanford University, has an MBA from the Wharton School of the University of Pennsylvania, and is an award-winning social innovation teacher.

About the Bessel Origin program

Bessel Origin is a real-world work-prep program for emerging professionals and people making a pivot professionally so that they can thrive on teams. We’re different from other courses or programs because of our unique combination of design thinking and agile teamwork, active mentorship, interactive workshops, and hands-on small team projects.  

For professionals, the Bessel Origin Program is a proven way to adopt best practices for working on a team and learn through real-world work experience. For employers, Bessel executes critical innovation projects and helps identify, develop, and retain star performers using evidence-based talent management.