Matlab is proprietary and requires a license, you may learn it but when you leave a place where they provide you with a license you will have to fork out for that yourself.

Python, R, Julia and many others (even the classic UNIX tools) are all Free Open Source Software that you can learn and take the knowledge with you anywhere, knowing that you will be able to use the tools in any setting.

It depends on your projects, really.

If you have to build upon or maintain existing (Matlab) code, then you don't have a choice.
Matlab is a commercial product and you have to buy a license to be able to even run or compile your code.

If you want to learn programming with a new project, Python is a good choice.
Python is open-source and always free and evolving. Code is backwards compatible and easy to run.

> who wants to learn programming (from scratch) to do data analysis

I think Python (and also R and Julia) is much more prevalent in the data science field than Matlab. There are certain engineering and applied math field, where Matlab probably still provide an ecosystem advantage, but I don't think data science is one of them.

Besides, Matlab is also not great on a lot of fronts. Being a proprietary language that requires a license is one, and its poor language design is another.

Sensitive content Reveal/hide

Programming? Python. Matlab is a complete language with a rich environment, but in some ways it's more designed like a maths software than a general purpose language. Things like basic algorithms and data structure can surely be taught and learnt decently using both, but if the goal is to learn programming in general, starting with Python will open infinitely more doors, and avoid a few of Matlab's limitations, quirks and bad habits (stuff like indexing starting with 1, etc.).

Besides, Matlab is on a downwards slope and has been for a while, even for its "traditional" uses. Even when I was in academia a few years ago, I sometimes had to work with Matlab, but it was either 1) large legacy projects everyone wished were rewritten in something else (often Python) but nobody had the ressources to or 2) the work of someone - generally not a computer scientist - who only knew Matlab. I can't recall anybody actually starting to create something and deliberately choose to use Matlab over other possible options.

I used MatLab at university and even wrote my dissertation project in it. In the 20 years since I've graduated I've not used it once. I've used R a bit, but mostly Python for my Data Analyst and Data Science jobs, in an engineering field.

My employer does have MatLab, but I think they're moving away from it, mostly because the licensing and version management (and the licensing of the different versions) is difficult and expensive. I've certainly not seen anyone start any projects in the last ten years with it.

Python versioning and environments is a bit of a pain, and I believe R is somewhat better, but tools like UV are making it better. If you're just starting out and using just the (many!) standard modules that come with Python, you can leave worrying about version management of external dependencies a little later.

I think for many businesses the MatLab simulation modules and extras are quite useful, especially if you're a smaller business, as they can be very industry specific and specialised - but cost extra again, and can get pretty expensive.

As an academic I recommend Julia. It is a super nice to write language with a great package manager (this really matters!). Add to this Makie for a great plotting experience.

TBH you will probably need some Python experience at some point because it is so common.

Matlab was actually my very first language. I haven't used it since I left academia though, whereas I've become a Python expert.

Packaging/install is absolutely the achilles heel of Python but it's otherwise a fantastic language and I think it's worth pushing through setup struggles. I also think the reach and community of Python both within and outside academia dwarfs Matlab and makes it the better choice even if the language itself is better to learn (a big 'if')

Matlab: better for numerical analysis, data handling etc.
Python: Not as good at the above (although still very useful), but with far broader applications.

If it's their first go at programming, Python, it's more widely applicable and it'll be easier to get to grips with Matlab if they know Python, rather than Python if they know Matlab.

Academia viewpoint here:
Go the python way.

Its also easy to learn and your students will benefit on the long run.

Had to use Matlab during study, but could never re-use it afterwards.
Self-trained python and use it now for a lot of things.
Data analysis, data visualisation.

I even created a package (qrcite) to automatically generate qr codes for a set of references, to use them in my presentation.

I used to be an academic but currently am not, so I voted as a non-academic, but idk 🤷

Anyway, I am definitely on the Python side because it's so much more widely used and easier to extend with packages. Plus it can do so much more than just numerical analysis. Almost every time I used Matlab in my days as an academic, I wished I could be using Python instead. It is true that Python can be less convenient, but I think the flexibility makes it worthwhile, and it's really not much more difficult if you learn how to work with it properly.

I would probably recommend Jupyter Lab as a frontend. Specifically: make a virtual environment (perhaps with pipx or uvx), install jupyter, install the extension that lets you pip install packages from within the Jupyter web UI, and that's all the messing around with venvs and CLI commands that you need to get started.