Where do you begin in machine learning?

Is actually breaking ground and getting started a question of motivation?

In this post discover the personal story of how finding the landmarks in machine learning made the difference for one software engineer.

Find The Landmarks in Machine Learning
Photo by Craig Stanfill, some rights reserved.

I received an inspiring email from Cliff Bryant in the new year.

Cliff is a senior software engineer looking to get started in machine learning. He has started his journey. In his email he was pointing out that it was not an issue of motivation, but an issue of finding landmarks helped him to get started.

I wanted to share his email (with his permission of course) because I think it is inspiring if you are still trying to figure out how to get started.

My initial problem with machine learning, and this went on for several years, was that I could not see the forest for the trees.

From the outside, machine learning looked like a bewildering collection of techniques to be learned. I could not see the structure and relationship between the different methods. I had no clue about which method should be applied in a particular problem, and how to attach a machine learning problem.

Now that I have a couple of machine learning online courses under my belt, I am beginning to see some general structure. E.g., supervised vs unsupervised learning, and regression vs classification.

I think the problems for newcomers to machine learning is “where to begin?” and “where to spend my time learning?” Having identified these landmarks, I now feel more confident in broadening my knowledge to include additional techniques.

Besides learning the algorithms, techniques, and theory, there are the practical issues of manipulating data at scale. It helps to have seen some examples in this area, or at least to have an overview of what is going on. In this respect, I view the R language as (in its current implementation) limited to modest sized data sets (that will fit in main memory). I see Python as a language that can span the range data scale problems.

It is helpful, once you begin to see the overall outlines of the machine learning field, since then it becomes easier to see how new techniques and algorithms fit into the overall framework.

My problem from the beginning was not one of motivation. I have always loved mathematics, from an early age. The problem that I had with getting started in machine learning was where to begin?

Great email! Thanks again for sharing Cliff.

If you have your own story on getting started, I’d love to hear it.

Leave a comment below or send me an email.