search

betanalpha / knitr_case_studies

Inference case studies in knitr
162 stars 94 forks source link

expression for prior_sd_lambda on inverse gamma is not quite correct #20

Closed arabidopsis closed 4 years ago

arabidopsis commented 4 years ago

There are 2 places in principled_bayesian_workflow.Rmd where the expression for the sd of an inverse gamma is used viz:

prior_sd_lambda <- sqrt( (9.21604)**2 / ((3.48681 - 1)**2 * (3.48681 - 1)) )

I think this should be (note the -2):

prior_sd_lambda <- sqrt( (9.21604)**2 / ((3.48681 - 1)**2 * (3.48681 - 2)) )

BTW: awesome work!

betanalpha commented 4 years ago

Thanks! There are numerous small errors, including the prior specifications and the prior standard deviations beyond the first iteration. I will be updating everything here, along with the addition of new material, once I finish a few more new case studies. I will be sure to rectify this along with everything else.