search

arviz-devs / Exploratory-Analysis-of-Bayesian-Models

Source repository for the online book Exploratory Analysis of Bayesian Models.
https://arviz-devs.github.io/Exploratory-Analysis-of-Bayesian-Models/
MIT License
17 stars 10 forks source link

Seeking more incremental feedback on Autocorr #1

Closed canyon289 closed 5 years ago

canyon289 commented 5 years ago

Per @ColCarroll's comments added an example of an autocorrelation plot that looks poor.

Have not included @ahartikainen's recommendation of interactivity or parts of a plot

Goal for this doc is to iterate a lot to get it to a high standard, then use this as the benchmark for the docs for the rest of the plots

canyon289 commented 5 years ago

I've created a PR proposing development rules for this repo. Once we decide on that, that next course of action for this PR will be more clear.

canyon289 commented 5 years ago

@ahartikainen asked somewhere if the ArviZ autocorr calculation was the same as the "manual" calculation in the docs. It is, here's some proof

arviz_autocorr = az.autocorr(mu_0.values)
np.isclose(arviz_autocorr, auto_corr, atol=.01)
autocorrtolerance
canyon289 commented 5 years ago

Marked as Alpha and added todo note to add neffective. Resolved feedback for excess code output by semicolon or assigning vars were appropriate Merging due to the above