I've been working on a package for quite some time, and as I get closer to a potential release, I have been thinking more about what to do next. And of course, just about the first thing would be to find out how many people have downloaded it.
We all know that RStudio has made it easier to get a sense of the relative number of downloads of different packages by aggregating information from its own many mirrors around the world, and that access to this data via the RStudio API is eased by a number of packages, etc. (e.g. cranlogs, adjustedcranlogs, & the shiny app cran-downloads). Am I correct in understanding that RStudio is aggregating only its own mirrors, and not all the CRAN mirrors?
If so, does anyone have any idea how total RStudio CRAN mirror downloads compare to total downloads from all mirrors? Or does anyone have a suggestion of how such information can be acquired? If we just knew the RStudio share, we could blow up the RStudio results to get some estimate of a packages global downloads. I think RStudio’s mirrors are sufficiently diverse in geography, etc., to represent a reasonable approximation to a random sample, at least for any packages that are not tightly tied to a geographic region.
Actually, geographic diversity in interest in a package suggests that it would be useful to have total non-RStudio CRAN mirror downloads by country.
I am assuming that the list of CRAN mirrors on CRAN is substantially complete. Does anyone have any reason to doubt this, aside from company-internal mirrors and the like?