Entirely depends on the project you want to build
post
Pandas. Python's only killer library imo
That's like saying "what's the best ingredients to learn cooking with?", firstly it all depends on what your want to eat, secondly it doesn't really matter what the ingredients are to learn cooking skills.
Flask is a fun framework for making web apis.
Pika is a client for RabbitMQ, if you want to try message queue stuff.
Numpy and sklearn for numerical and machine learning stuff.
Matplotlib for making nice plots of your numerical stuff.
Pytorch for deep learning.
Pillow for image processing.
OpenCV for computer vision.
Pygame for 2D games (maybe a bit old, but I had lots it fun with it when I started learning programming years ago)
Standard library of course. And collections module too, among the others that come by default. What you need is assignments to solve. Like project euler or hacker rank.
If you want to do web requests/ use API's, use 'requests'
graphs/reporting, I've used 'bokeh' before, it was nice.
I've never used PyDroid, so I'm not sure how you'd install things, but these are both available via pypi, python's package repository.
Thanks but Im just going to go ahead and do my own thing. Everyone's very insistent about doing things a certain way in this thread and honestly its unsolicited advice. Not everything is an XY problem :). I'm just going to install my different, unconnected libraries and tinker in peace
I heard there are quantum computing libraries in Python if that interests you!
If I were you I'd browse PyPi for any packages that look cool.

all 22 comments