Javascript for DataScience

as you all know FCC is about javascript for web development (MEAN stack?) and is ok, i love it, but today the world of data is also important and more or less related to web…

I would like to know if there are frameworks or libraries to peform data science analyses with JS (statistics, machine learning, etc), and if so, ask to the FCC staff to perform a new certicate to learn the basics of this field…

if not, the same but with python :slight_smile:

like this?


MEAN instead is Angular as framework

You really want to use Python over Javascript for mathematical/scientific heavy applications, in general.

1 Like

I’ll second the python recommendations – it’s truly an excellent environment for statistics and machine learning given all the libraries and tooling around it, and hey, FCC has a course on it. You’re not without options in the JS world however (sorry for the Medium link):

1 Like

There are libraries (linked above) to do data science with JavaScript however JS is not the best language to do any serious work, due to 2 reasons:

  1. JS is not a readable or easy to understand language. Data science is a specialized skill in itself, meaning less time to focus on the code itself, and due to JS’s nature as being the language of the web, it has a lot of quirks and legacy choices built in for better or worse. In comparison Python has 2 significant versions (2, 3) where 3 is being more or less standardize (finally). This, along with the fact Python provides easy to read and understand syntax, while still being very flexible, means its easier to program with its syntax, and thus more time can be devoted to the use-case. This isn’t only true for data science, but it is one of the main driving forces behind Pythons recent surges in popularity

  2. JS is not a performance based language when it comes to CPU focused problems. JS is a single threaded language that relies on async event loop making it great for IO intensive tasks, but due to being single threaded there is a ceiling where multi-threaded implementations can “solve” problems faster. Don’t get me wrong JS is still fast, but due to its architecture it has its limits.

Python isn’t known for being fast either, how-ever, it is more flexible in its architecture to handle specific use-cases like multi-threaded environments, which might or might not be a critical pain point, but could be for data science use-cases.

Finally, if your goal is to get into data science, Python is the name of the game. You can learn using JS, but its second fiddle to Python in the data science realm, and I don’t see that changing any time soon.

1 Like

The readability of a language is an extremely subjective metric, and several of JS’s numerical libraries link to the same BLAS/LAPACK libraries that make numpy fast. I don’t think we need to get into language wars here: the landscape is wide open, and new options appear every day.

1 Like

Since it was brought up, I will highly recommend you use Python 3 over Python 2. Python 2 has been EOLed.

1 Like

Here’s a site that had this in mind. It makes use of a mix of base JavaScript, D3.js and lodash.

So it’s very powerful already! The one obstacle you’ll have to get over is that you are manipulating JSON objects and not rectangular data frames or Excel spreadsheets. This might be a bit odd to wrap your mind around it. Good luck!

i not heavy into JS, but i… dont know why you would want to force it for this sort of application over the aforementioned Python. i mean… we have soooo many libraries to tap into that make data work a breeze once you know them. most popular ones like numpy, pandas, scikitlearn, matplotlib, and so on… they’re just great and take seconds to pip install and run.

1 Like