Hi ,

​

In most of the world's countries, right-hand driving is the law. Yet, in approximately one-third of all countries, the opposite is true.

Knowing which side of the road to drive on is essential knowledge if you want to drive somewhere without a crash. But, on its own, a statement to the effect of "all drivers must drive on the left" is worthless, without the additional context of the countries to which it applies.

Here's the thing...

Every machine learning model is just a complex set of rules derived from previously observed patterns in data. But without the context of when those rules should apply, relying on a machine learning model can be as dangerous as driving on the wrong side of the road.

In Drew Conway's Venn diagram of what it takes to be a data scientist, machine learning sits at the intersection of Hacking Skills and Mathematical Knowledge. However, it is through the addition of Substantive Expertise (or context) that machine learning is elevated to data science and value is produced.

Machine learning is a highly useful skill and a necessary part of most data science roles. Yet it is the combination of machine learning plus context that produces value. Not machine learning in and of itself.

Talk again soon,

Dr Genevieve Hayes.

All past emails are available in the archive.