Is Javascript the language that will change DS/ML forever?

Introduction

Back in 2018, Tensorflow introduced a JavaScript interface to their list of supported languages. I think this came as quite a surprise to a lot of the Data Science community, but the decision certainly makes sense. JavaScript is a very popular programming language, and surely those who develop in it want to take advantage of what the Tensorflow library has to offer. Of course, machine-learning is not a very application of JavaScript at all, so there is a question to ask on why exactly this implementation occurred.

What is JavaScript’s place in the machine-learning ecosystem? Is there a chance that we might find ourselves writing JavaScript over Python in the coming years? These are just some of the questions I wanted to touch on and try to start a conversation about in this article. In Data Science, it is always important to be up to date on the latest technology in the field, and even though it might not seem like it, JavaScript is not an exception to that rule. So I want to discuss some of the short-comings of using JavaScript for machine-learning, and also discuss its potential impact on Data Science as a whole.

JavaScript’s Potential

JavaScript is well-known for its abilities in web-development, but how would those abilities carry over into an application in Data Science? First and foremost, we should consider that JavaScript is not a statistical programming language. The language was pretty much created for web-development, which the language excels at. This means that JavaScript from a typical Statistician’s point of view might be a bit hard to grasp. However, this also used to be the case with Python. Python itself was never intended to lead the world in machine-learning, but took the spot thanks to just being a great language in general. Is there a chance that JavaScript could take this position?

JavaScript certainly has potential, but there are some key things to consider about the language before we consider it a runner-up in the machine-learning space. The first thing is that JavaScript has a great ecosystem… for web-development. While Python has a lot of great packages for handling data, linear algebra, and machine-learning, this is simply not the case with JavaScript. It would take a long time for JavaScript to catch up in that regard.

Another thing to consider is that JavaScript is slow. Like Python, JavaScript is an interpreted programming language, and that comes with a set of short-comings. However, when it comes to languages that can use C for most of the back-end code, we see that often their speed does not tend to matter as much as it might in other languages. That in mind, ultimately while you might lose some speed by going from Python to JavaScript, in the example of Tensorflow, you might end up running the same exact code ultimately.

#javascript #data-science #programming #python