Publish the equation used to calculate the risk score

[if62668]

I see all documentation on the metrics used, but cannot see how these are used to calculate the 0-1 risk score for a package.
Can you publish this in "plain text", not R code.

[dgkf]

Thanks for raising this @if62668.

Yeah, that's a nice suggestion - we certainly agree with the sentiment here. The scores have been a long-standing friction point with this package. This scoring algorithm was never intended to be a de facto standard. It was implemented really just as an example, with the goal of making it flexible enough to accommodate a more thoughtful scoring process.

Somewhere along the line, that sentiment was a little lost and we found that most folks in the industry are happy with the default.

We've actually kicked off a series of successor packages since then and one of our goals has been to remove our scores entirely. The functionality is split between a package called val.meter and another called val.criterion for the assessment and scoring respectively.

In val.criterion we aim to be significantly more transparent about the scoring process. Scores in val.criterion can be more than just weighted averages, and we intend to represent them visually as a decision tree.

[if62668]

Thanks: for context I am reviewing various risk scoring practices proposed for R packages. SCTO have a very similar scoring methodology, and the scoring calculation is published here.
When I compare the scores for packages (same versions) between SCTO and PharmaR, the SCTO is consistently lower than PharmaR. Examples (pkg version PharmaRScore SCTOScore):

• mvtnorm.v1.3-3...0.41....0.17
• stringr.......v1.5.1....0.29....0.08
• survival.....v3.8-3...0.45....0.13

If we could see the calculation behind the PharmaR score, it may be possible to determine the reason for the difference.