Closed jmh530 closed 2 years ago
Merging #60 (6349867) into master (f61a82f) will increase coverage by
0.00%
. The diff coverage is100.00%
.
@@ Coverage Diff @@
## master #60 +/- ##
========================================
Coverage 99.92% 99.93%
========================================
Files 21 22 +1
Lines 2804 2930 +126
========================================
+ Hits 2802 2928 +126
Misses 2 2
Impacted Files | Coverage Δ | |
---|---|---|
source/mir/stat/distribution/negative_binomial.d | 100.00% <100.00%> (ø) |
Help us with your feedback. Take ten seconds to tell us how you rate us.
negativeBinomialInvCDF
is a little time-consuming for extreme values in a naive implementation, even with a good guess. I took some inspiration from R's version of it to get it working in a more efficient manner for some extreme values, but it's not exactly the same. A simple copy of R's version also wouldn't pass the unittests, I think, even if I wanted it to. Some adjustments versus R's version enable it to pass UTs here.