[BUG] baccuracy()-function returns NA 🤦♀️
#24
Assignees
Labels
bug
Something isn't working
Comments
serkor1
added a commit
that referenced
this issue
Dec 21, 2024
* Added `na.rm`-argument that divides and adjust by number of valid classes if TRUE. * Adjusted by (1-p) and not (p) * Updated `baccuracy()`-functiotn documentation * Fixed a bug in the unit-test for `baccuracy()` where adjust flags werent passed correctly.
serkor1
added a commit
that referenced
this issue
Dec 21, 2024
* [BUG-FIX] See issue #23 🔨 * Fixed a bug in `fbeta()`-unit test where `micro`-flags werent passed correctly. * Fixed a bug in `fbeta()`-function where `micro`-flags werent passed correctly to header file. * Simplified F Beta Score calculation * Switched from np.nan to 0 in {scikit-learn} implementation. np.nan were too inconsistent. * Replaced all NaNs with 0 in the unit-tests * [BUG-FIX] See Issue #24 * Added `na.rm`-argument that divides and adjust by number of valid classes if TRUE. * Adjusted by (1-p) and not (p) * Updated `baccuracy()`-functiotn documentation * Fixed a bug in the unit-test for `baccuracy()` where adjust flags werent passed correctly. * [BUG-FIX] `RRMSE()`-function replaced 🚑 * This function was supposed to be the Relative Root Squared Error. See https://www.gepsoft.com/GeneXproTools/AnalysesAndComputations/MeasuresOfFit/RootRelativeSquaredError.htm * [BUG-FIX] See commit message 📚 * Unit-tests for generic classification functions werent passing micro flags correctly. These have been fixed. * The `rae()`-function was calculting means, but it should have calculated and compared sums. This has been fixed * Update documentation for `rrse()` * Switched expectation function for matrices in S3 unit-tests; as.table removes all 0 entries so set_equal cant be used. * [DOCUMENTATION] NEWS updated 📚
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
issue
In imbalanced classification problems
baccuracy()-function can potentially returnNAsource of bug
https://github.com/serkor1/SLmetrics/blob/a90db17b48fcbae74a857afe5b0f53faaf7a1f97/src/classification_BalancedAccuracy.h#L29C9-L43C70
If one of the rows are 0, the entire metric is going to be
NAfix
A similar solution to this:
https://github.com/serkor1/SLmetrics/blob/a90db17b48fcbae74a857afe5b0f53faaf7a1f97/src/classification_Helpers.h#L39C1-L55C2
The text was updated successfully, but these errors were encountered: