The Myth of the Normal Curve and What to Do About It - Allan Campopiano | SciPy 2022

📊 Interactive notebook app:
deepnote.com/workspace/allan-...
🐍 Hypothesize (robust statistics library for Python):
github.com/Alcampopiano/hypot...
✍️ Proceedings manuscript:
conference.scipy.org/proceedi...

[@AlCampopiano]

For anyone looking for the resources shared in this talk, here you go:
📊interactive notebook:
deepnote.com/workspace/allan-campopiano-4ca00e1d-f4d4-44a2-bcfe-b2a17a031bc6/project/robust-stats-simulator-7c7b8650-9f18-4df2-80be-e84ce201a2ff/%2Fnotebook.ipynb
🐍Python library for robust stats:
github.com/Alcampopiano/hypothesize
✍Paper:
conference.scipy.org/proceedings/scipy2022/pdfs/allan_campopiano.pdf

[@fburton8]

Q: How many data points are needed to tell (to some level of confidence) if they are drawn from a normal distribution or not? Is there a good procedure to answer this question?

[@wwymak]

Really useful talk-- any chance you can share the link to the notebook here?

[@AlCampopiano]

Now added to the description! ❤

[@fburton8]

Excellent and thought provoking talk. 10:42 Is standard deviation really the same as standard error?

[@AlCampopiano]

Thanks! Yeah, that was a tiny slip up on my part. They are both measures of variability. SD reflects variability in a sample. SE reflects variability across multiple samples from the population. In the context of my notebook, I'm calling np.std() on a sampling distribution of means, hence my tiny slip up 🙂🙃!

[@fburton8]

@@AlCampopiano That makes sense, thank you!

[@MatthewBrett]

Yes, the standard error is the standard deviation *of the distribution of sample means*.

[@AlCampopiano]

@@MatthewBrett Great to see you here Matt!

[@MatthewBrett]

@@AlCampopiano - thanks for the talk. I went off to find a copy of "Fundamentals of Modern Statistical Methods".