In November 2019 I gave an invited tutorial on Probability and Statistics at PyData NYC. I had a great time putting this talk together and it has given me a bunch of new ideas on how I want to think about probability and statistics going forward (and I was also really inspired by many of the other great presenters there)!

Very excited I got to make a few points about the mind projection fallacy!

The talk ambitiously aims to cover all of statistics in 90 minutes. The main idea is to start from probability, cover parameter estimation and hypothesis testing and then cover how the generalized linear model can be used to extend and expand this. I’m pretty happy with the results. Follow are is the video for the talk, as well as a link to the github repo containing all the notebooks used. As some bonus content there is an implementation of probability as logic using Python’s Data Model methods that I really wish I had thought of before this presentation. I had fun giving this talk so I hope you have fun watching!

The Video

Talk Code and Notebooks:

https://github.com/willkurt/ProbAndStats-PyDataNYC2019

Bonus: Probability as Logic revisited

In the first part of the talk I go over how we can view probability as logic. Only after this talk did I start thinking of how I could implement negation, or and and using Python’s data model methods to create a simple probability class that showed how the logical operations can be mapped directly to logic. This jupyter notebook on implementing probability as logic is a really fun way to see how we really can use probability as a continuous form of logic.