The Most Important Object In Python: A Full Tutorial

Check out my course on UDEMY: learn the skills you need for coding in STEM:
www.udemy.com/course/python-s...
Code from this video: github.com/lukepolson/youtube...
Join the Discord Server: / discord
Follow on Instagram: / mrpsolver
0:00 Introduction
0:49 Basic 1D Arrays
3:05 Intro To Multi Dimensional Arrays
10:32 Array Reshaping
12:23 Broadcasting Rules
16:51 Broadcasting Example
19:04 Operating Along Dimensions
23:05 3 Practical Examples

[@wanadzhar3031]

This isn't a shitpost.

[@4001Jester]

i was honestly hoping for the bumpy array

[@MrPSolver]

You know what, maybe I will make this anyways.

[@nyden94]

One of the best numpy videos I've found. Every time I get confused with dimensions, I come back here!

[@buhariibrahim4055]

@Mr. P Solver, I am an engineering grad student without coding background from undergrad. For some time now, I have been trying to learn programming for a while now jumping from tutorial to tutorial and just recently stumbled upon your channel. I must say, you are by far the best I have come across. Your examples are targeted and straight to the point, and are also very relevant to real numerical computation problems. I therefore want to express my gratitude for the time and effort you spend in creating these videos. Thank you!

[@MrPSolver]

Cheers :)

[@fransdebruijn6763]

@@MrPSolver The quality and placement of the microphone makes a HUGE difference. many tutorials are some Indian dude with a scratchy microphone in an room with a lot of echo. Others are with a crappy distorted compressed headset microphone. completely unusable.

[@calebstamper1419]

Similar boat and strongly agree!

[@ronmurayama4469]

Excellent explanation of making complex procedures simple, thank you Luke!

[@urban_sculptor]

I previously found broadcasting really confusing, but now all makes sense - thanks a lot!

[@cianjones2728]

Very underrated chanel, good vid

[@TheTWAlbert]

Bro this is a nice one, you just clarify the concept of numpy so clean. Nice job!

[@AnimeYTube]

My man, i get in this channel by Billy, but i found one of the best tutorials that i've ever see, u open my mind, thank you and continue, when i star earning money remember me to donate something to you haha

[@andrewglick6279]

Fantastic video! It effectively sums up (pun intended) everything about Numpy arrays in an organized manner. I will definitely be recommending this to people. One suggestion I would make is to not use images of size 256 in the future. The numbers 255 or 256 in the context of images usually makes me think of pixel values rather than size, so I had a bit of confusion there for a bit. A size of 128 or 512 would be fine, though.

[@algiffaryriony]

Great video! I hope more people learning ML and using numpy get to see this.

[@AWest-ns3dl]

Dimension analysis is also great for making sure your units work out in an equation. E.g. 6m / 3s = 2m/s

[@charlesspringer4709]

You gotta remember to leave an empty 10 seconds on the end so we can see how it finishes without the thumbnails in the way.

[@hoonji126]

Sir this is absolute gold. Please turn on super thanks so I can express my gratitude in usd dollars.

[@sagsolyukariasagi]

Great content! Thanks for this Luke. Can you please prepare a video tutorial about physics informed NN's?

[@random_netizen]

O.M.G this ia what i need to complate my final project, huge respect

[@TaifPatwary]

Yo this is great man!! Thanks for this content

[@Piipolinoo]

Awesome video, as always! Keep it up :)

[@prietjepruck]

Thanks for this great tutorial. For me this is the best way to learn. Keep up the good work.

[@philippe6724]

Amazing ‘to the point’ lesson

[@karansmittal]

This was really helpful!!

[@khidhrhalab3543]

thanks for the knowledge you spread !, i have a question , on the last example ,isn't the green channel index supposed to be 1 rather than 2?

[@deepgalxy]

Really love your video, super good

[@sonugeorgealex]

Thank you for such a wonderful lecture🥰🥰

[@AJMansfield1]

26:00 This is not an appropriate or correct use of `reshape`; it's incredibly brittle and breaks with any change to the shape of the input. The function you actually want there is `expand_dims`; or, better yet, pass `keepdims=True` as an extra argument to `sum`. Doing it with expand/keepdims means not only does it not break when x becomes a (4,50) array instead of a (3,50) array, it even still works when you add on another dimension to make it a (4,50,3) array.

[@harrytsang1501]

Numpy is great. Some courses that used to use R and Matlab are switching to it. Being accessible aside, numpy is essentially just compiled C with python bindings. That's why it can be fast and the same api can be used to take advantage of CUDA hardware with little overhead from python.

[@aflous]

Nice one Mr. P ;)

[@einsteinsboi]

Fantastic video thank you!

[@estrastyven3752]

Thank you shregory!

[@iStrawHat07]

I literally have no idea what the hell is Numpy but I watched this video anyways. Liked and subbed too. Hah!

[@houstonbova3136]

Have you checked out polars at all yet? It’s a lot quicker than pandas.

[@mindeyi]

I'd like you to cover xarray library ^__^

[@sunchimoonchi]

I'm that Anything!!!. Just learning...lol

[@degenyakuza]

the magician

[@subhashsingh8553]

Sir, pls make video on geometry also-circles,traingles,quad on python

[@tilkesh]

Thanks

[@tomquareme3787]

Numpy is realy nice if you can vectorize your code or can use list and generator comprehensions. But sometimes when I can't vectorize and have to resort to for-loops I use Numba. It's fast and great and all, but when you have to do something non-trivial involving dates and np.datetime64, it's hard to avoid cryptic errors and I tend to fall back to object-mode for some parts which will be slower... I really wish in the near future the Numba subset of Python will be expanded and most of the standard lib stuff will be usable. I was thinking to switch to Julia, but I don't know... It's ecosystem of libs for data science and ML is a lot smaller, JIT compilation can be slow for short lived programs, 1-based indexing is a bit meh but not troublesome though. Sometimes I think C/C++ isn't that bad at all, but Python's convenience for libraries and string manipulation is hard to beat. Feels like I'm stuck between suboptimal solutions.

[@AWest-ns3dl]

My rabbit hole of optimisation:. native, numpy, patience, numba, jit, cython, patience, dask, spark, fortran.

[@stymlice2332]

@@AWest-ns3dl how tf you even go from python to fortran 💀

[@AWest-ns3dl]

@@stymlice2332 after a double serving of patience.

[@TheGmr140]

Numpy is magic

[@benheet6730]

Which vs code theme is it?

[@shikharrana2229]

It's jupyter notebook on Vs code

[@benheet6730]

@@shikharrana2229 yea I know, talking bout that green highlighting of the font

[@Zzznmop]

Would the underlying data structure be more import, since numpy arrays can’t exist without them?

[@I_Unintentionally_Morph]

Wow

[@SOLOMON-ALEXANDER]

Python is easy😆💯🔥that's why it's my favourite language... I learnt this numpy... Import numpy as np👍

[@erichstocker8358]

Numpy is NOT a part of basic python. While the numpy module is a great module and is definitely part of the anaconda distribution (a very good one for science and engineers), numpy should not be presented as if it were part of basic python. You should state that using the addon numpy array is a key technique for people who use that extension; not that it is part of python.

[@chinchillamdgamer]

Wait ... Wheres the parrot?

[@user-jq5bz4wp2e]

All that is well and good. But why can't you do a reverse sort on numpy arrays? I can get it done but I have to jump through hoops. For example you can multiply it by -1, sort it and multiply it by -1 again. But why?!? Sometimes I wake up at night screaming "WHERE IS MY REVERSE SORT!?!?". It haunts me. It would be so simple to implement. I can't help but think they are just screwing with us. Or maybe just me. I can't sleep. I can't eat. I can't think about anything else. My life is ruined. Bastards.

[@nehemiah6280]

He makes real coding videos?

[@rallokkcaz]

Except numpy isn't part of Python. It's a library.

[@aatkarelse8218]

Wrong, the best skill is structuring your code so that others can still read it, lots of so called data scientists fail at this.

[@MrPSolver]

Quite true!

[@Eeeff]

If you're the only DS on your team, then why would it matter how readable your code is? Surely other skills are more important in this case.

[@aatkarelse8218]

@@Eeeff And you plan to stay as the only DS in your team?

[@Eeeff]

@@aatkarelse8218 No, but in the scenario where you are the only DS, having organized code comes second to functionality.

[@aatkarelse8218]

​@@Eeeff So you refactor your code after it is done, as in it functions?

[@TheDarkhawk243]

stop waving your hand when they dont need to be waved lol