Rather than knowing intrinsically what parameters are required for a distribution, scientists tend to have a sense of what value they expect a measure to take, how many observations should fall within a certain distance of that value.
For the normal distribution, this is straightforward, as the parameters reflect the expected value and variance. However, for the beta distribution, the parameters are not so readily interpretable.
parameterpal:: provides a means of obtaining the parameters required for the beta distribution from interpretable conditions.
devtools::install_github("softloud/parameterpal", build_vignettes = TRUE)
After installation, the
parameterpal:: tutorial will be available in the Tutorial pane of Rstudio.
# executing this code will launch the tutorial in your browser learnr::run_tutorial("ppalhelp", package = "parameterpal")
This package was developed for a friend and collaborator, computational ecologist Dr Matthew Grainger, who previously used browser-based tools to obtain beta parameters to inform his rstats workflow. I hope he finds this package a useful augment to his codeflow.
more information the same information, but from the handy ease of your Rstudio Viewer pane.
Full disclosure, I need to update it after I fully understand this gist I was provided with after posting the first version of this software to twitter. Open science at its best. Blogpost incoming on open science and why it’s good to share bad math and beta research code. So cool I’ve ended up with a better, more mathematically correct solution after posting it.
No reason the same math can’t be applied to other distributions. Open an issue if you’d like me to provide parameters from interpretable conditions for another distribution.