Closed s-weil closed 8 months ago
Note that in a separate issue, one could try to tackle code duplications in the src and tests of distance metrics. This however may affect performance and must be conducted carefully.
Regarding that p could also be 0 < p < 1
and restricting it to be an integer, I may have been a bit too strict on the given signature specified in https://github.com/fslaborg/FSharp.Stats/issues/296
Also it was specified to be an integer in https://en.wikipedia.org/wiki/Minkowski_distance
Happy to change it to a float for a more generic version. It may affect performance though.
If it becomes a float, I can also handle the case when 0 < p <1
by removing the exponent 1/p
as described in Wiki.
Pushed the feedback.
The order p
is now a float and the case 0 < p < 1
is covered
Merging #308 (9d38186) into developer (1e680b1) will increase coverage by
0.27%
. Report is 12 commits behind head on developer. The diff coverage is79.57%
.
@@ Coverage Diff @@
## developer #308 +/- ##
=============================================
+ Coverage 46.73% 47.00% +0.27%
=============================================
Files 148 148
Lines 16225 16458 +233
Branches 2196 2219 +23
=============================================
+ Hits 7582 7736 +154
- Misses 7987 8052 +65
- Partials 656 670 +14
Files | Coverage Δ | |
---|---|---|
tests/FSharp.Stats.Tests/DistanceMetrics.fs | 100.00% <100.00%> (ø) |
|
src/FSharp.Stats/DistanceMetrics.fs | 25.23% <43.52%> (-15.68%) |
:arrow_down: |
Please reference the issue(s) this PR is related to
Closes #182
Please list the changes introduced in this PR
DistanceMetrics.fs
for vectors, arrays and sequencesDescription
Implement the Minkowski distance of order p in several versions, following https://en.wikipedia.org/wiki/Minkowski_distance and #182
[Required] please make sure you checked that
[Optional]