written by Eric J. Ma on 2016-08-18
Recently found a very good article on the ArXiv about Bayesian estimation for difference in sample means. Loved the fact that someone's done the work so that I don't have to!
The paper essentially explores how a Bayesian estimation (BEST) is essentially much, much better than traditional Null Hypothesis Significance Testing (NHST), and the author provides a good number of convincing arguments for this. The author basically covers the most commonly-used case in the experimental sciences - inter-comparison between an experimental and control group. I'd recommend reading the paper (link).
While skimming through it, I then found a PyMC3 recipe, as usual, provided by the awesome Thomas Wiecki and Chris Fonnesbeck.
It makes me wonder... given how Bayesian inference is much more natural than Frequentist inference, why isn't this used more often?
I send out a newsletter with tips and tools for data scientists. Come check it out at Substack.
If you would like to sponsor the coffee that goes into making my posts, please consider GitHub Sponsors!
Finally, I do free 30-minute GenAI strategy calls for organizations who are seeking guidance on how to best leverage this technology. Consider booking a call on Calendly if you're interested!