Closed AndreaPi closed 6 years ago
We don't know which download is direct and which one is not.
As for the comparison, the UI for that is a bit hard to write, I think it is simpler to write an R package for it.
@gaborcsardi thanks for the reply.
Concerning the direct/dependent downloads, there must be a way to find out, though I have no idea how difficult it would be to implement it. If you look at https://www.rdocumentation.org/packages/glue/versions/1.2.0 for example and hover over the "i" next to the number of monthly downloads, you see the following tooltip: As a matter of fact, the API of RDocumentation allows splitting the downloads in direct/indirect: https://www.rdocumentation.org/docs/#api-Package-Get_splitted_package_downloads Maybe a simple solution could be to query their API?
Concerning the UI, no worries, if that's complicated and/or time consuming please just forget I mentioned it.
They have a statistical model for the direct and indirect numbers, but in my experience it is not very good, the numbers do not seem plausible at all.
Even their totals do not add up to the correct numbers.
I didn't notice that! Then it's not a good idea.
I think a useful addition to the METACRAN site would a ranking of packages by direct downloads. Most downloaded seems to be based on total downloads, i.e., including dependency downloads. No surprise
Rcpp
comes first - it's also one of the most depended upon CRAN packages.Also, it would be nice to offer the possibility of comparing two or three packages one against the other - i.e., add a form where I select, say,
skimr
andsummarytools
, and get a table which compares them by number of direct downloads, downloads this month, stars on GitHub, etc.