The “sample” function in the statistical programming language R is biased… See screenshot.
Possible fix: github.com/daqana/dqsample
1
12
2
30
MacOSX R from here: cran.r-project.org/bin/macos… Bug seems inexcusable. I’ve recommended R over python when correctness is critical, because R’s vetted by statisticians not programmers. Made a fool of me?
3
1
The bug was reported + fixed before R 3.6.0, see the list archives, or of course the R sources.
As I told @lemire over DM (which I prefer over yelling here) his mistake may be the restoration of prior session settings. Try pristine R as I did with the Docker example. No bug.
Dec 27, 2019 · 11:00 PM UTC
2
1