Why did machine learning resurface again

Last Updated on

Discover Your Personal Why And
Finally Get Unstuck

In this post, we will explore why you are interested in machine learning.

We will look at some questions that can help you get to the root of what draws you to the field.

We will finish with a map showing the 4 main “whys” so that you identify where you fit and what resources to target.

Question Your Why

Why are you interested in machine learning? Have you deeply considered this question?

It is useful to know your why, because you can use it as a filter to best choose the projects and tasks that you enjoy to work on. If you cannot come up with a clear why, that can be useful too as it can motivate you to try a bunch of different things and find out what you like or want to do.

You may be drawn to machine learning for lots of reasons. Perhaps you are responding to media and news articles about big data and data science. Perhaps you have seen a glimpse of machine learning in a tool or from a friend and you think it’s cool. There may be many reasons, but learning machine learning is hard work. To have the confidence and persistence to get through studying the hard and frustrating parts, you will want to have a strong why to fall back on.

I’m going to pose some rhetorical questions, and I want you to think about (even write down) your answers and see which one gels with you the most. One question is not better than another – keep an open mind.

What do you want machine learning to do for you?

Solve a Problem

Do you have a problem that you think machine learning can solve?

Maybe it’s an open business problem or a problem at work. Maybe it’s an opportunity you can see in the market. Nevertheless, you are thinking about machine learning as a tool for you to learn and apply to a problem.

In this case you may be interested to learn tools that provide implementations of algorithms you could use quickly. You will also very likely be interested in the creative ways to use these tools, such as case studies on problems like the problem you want to solve.

Technical Achievement

Is learning machine learning a mark of achievement?

Maybe machine learning is a popular technical field and you get great pride from learning new and difficult technologies and tools. Maybe you see machine learning as your next big challenge and opportunity for growth and a chance to demonstrate your abilities to learn and master technical materials.

If this sounds like you, you may be interested in books of algorithms where you can fast track getting an understanding of a method and how to use it without having to get down into the latest research. You will also very likely be interested in completing courses, entering competitions and implementing algorithms yourself.

What do you want to do with machine learning?

Extend the Field

Do you already have some experience with machine learning and want to extend what is possible?

Maybe you have been around the block with machine learning and read a book or completed a course. You have found a question or a method that you just can’t put down and not only do you want to go deep on that method, but you want to push the boundaries of what that method can do and has been shown to be capable of.

If this rings a bell, you may be interested in deep subject matter on the subject such as research papers and monographs. You may also be very interested in hearing expert opinions on the subject and exactly where the edges of the frontier are.

Do What Was Impossible Before

You have some experience with machine learning and you have some domain expertise and you want to do things in your domain that are not possible without machine learning. These are not necessarily problems like those mentioned above in the “Solved Problem” section, but rather the extension of a domain using experience of and capabilities provided by machine learning.

You will be interested in methodologies from data mining to automatic discovery of patterns. You will also very likely be interested in case studies of discoveries and extensions made by machine learning methods in similar domains.

Machine Learning Map

This is all a gross simplification of the field, but we could classify the motivation to learn machine learning by the type of work we want to do. We can classify the type of work we want to do into solving a problem in machine learning or in another domain. You can classify the types of tasks as tasks of a practitioner and tasks of a researcher.

I have tried to capture this summary in a table, see below.

The table has two rows by domain: the domain of machine learning and the other domain (such as analytical chemistry, petroleum mining or transport analysis.). The table has two columns by role: practitioner and researcher. Each box has the type of task for that domain-role intersection which is either solve a problem or extend the field. And each cell in the table lists the types of resources that may be of interest to a person interested in that task.

Map showing the 4 why’s for getting into machine learning.

Each cell can be considered a why that is motivating you to learn more about machine learning and the list of resources are things that can help in that pursuit.

This is just one way to slice the pie, but I’ve been meditating on it for a few weeks now. I worked hard on the groupings and I’m very interested to hear what you think of it, please leave a comment. I’d love to get some pro’s to start poking holes in it so we can see the strengths and limitations of this model (all models are wrong, it’s just a matter of degree).

Please leave a comment and let me know where your why fits in and what you identify with.

I have to say thanks to my wife for helping me think through this and map it all out on a whiteboard.