Hackers Guide
5 Probability Distributions Every Data Scientist Should Know
Thomas Wiecki - Probablistic Programming Data Science with PyMC3
Vincent D Warmerdam - The Duct Tape of Heroes Bayesian statistics
pymc tutorial - http://pymc-devs.github.io/pymc/tutorial.html
pymc examples
https://www.users.csbsju.edu/~mgass/robert.pdf
https://www.slideshare.net/CorrieBartelheimer/bayesian-workflow-with-pymc3-and-arviz
Frequentism and Bayesianism: What's the Big Deal? - SciPy 2014 - Jake VanderPlas
Frequentist vs Bayesian vs ML (triangle)
[history of MCMC tools]
Picture at min 6:30
- frequentist -> Bayesian
- hypothesis testing -> estimaiton with uncertainty
Example
- coal mining data 1851 - 1962
min 8 - picture of Bayes theorem
post = likeil * prior / model evidence
non-informative prior
Nusiance params (min 10)
Incorrect prior = biased results
Uncertantity
- freq = if this experiment repeated many times, in 95% of the cases the computed confidence intervals will contain the true parameter
- bayes = given the data, there is a 95% probability that the vaule of the paremeter lies in the credible region
freq = vary the confidence interval, fixes the param bayes = varys the param, fixes the credible region
example at 1820
bayes = Probabilistic statement about parameters given a gixed credible region
Freq = Probabilistic statement about a procedure for generating confidence intervals given a fixed model parameter
pic on min 21