koaning
commented
4 years ago Hi Kamil,
thanks for the feedback! I've been thinking about this theme of "combining topics" for a while. This is also why there's a few videos on datasets; the idea is to solve a practical problem and to show how I might go about it. Here's a list of things on my mind:
- making a visualisation service using altair and fastapi
- writing unit tests for pandas pipelines
- how to use cvxpy to write your own scikit-learn estimator
I've found sofar that the dataset videos are the least popular though. And there's also still so many tools that really deserve attention. So a part of me likes the idea of making more of this "combining tools and solve a problem" but it's not high on the priority list.
I want to be little bit careful with generative art. I certainly agree that it is a wonderful topic but I think I'd prefer calmcode to stick to being a place where you can find "tools and thoughts that might make your professional life more enjoyable". I don't know if generative art is something that fits the professional life and there's a huge backlog of ai-fairness/pandas/scikit-learn related material that I think deserves priority at this phase.
If you know of any tricky datasets, possibly where fairness is an issue, I'd be all ears though.
kamil-kaczmarek
Hey @koaning,
First of all thank you for making calmcode - I really like the videos and selection of tools presented :)
The idea You present number of tools and show video-tutorials how to use them, starting from scratch. I think that it might be also valuable to show working example of few libs put together in some mini-project. Such project should be about some particular topic or solve some problem.
I have in mind one particular idea: Generator of the images that looks like Piet Mondrian's abstract paintings.
Example paintings: one and two. Many more examples here.
Why I think it's interesting? There are few arguments:
Please let me know what do you think about it?
Best, Kamil