You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
In umap_metrics.py module many functions can be found with very similar purpose and naming.
For example gower_dist_numerical_old and gower_dist_numerical are different implementations meant to compute Gower distance between two vectors.
Similarly the module contains tanimoto_gower, tanimoto_gower_old and tanimoto_gower_no_bool_weights.
My proposal is to accurately check the appropriate algorithm that can compute tanimoto-gower distance in an efficient way, and verify its results with tests in test_metrics.py.
HINT: tanimoto_gower and gower_dist_numerical do not seem to be working perfectly, but according to previous comparisons, they seem to lead to better results.
The text was updated successfully, but these errors were encountered:
In
umap_metrics.py
module many functions can be found with very similar purpose and naming.For example
gower_dist_numerical_old
andgower_dist_numerical
are different implementations meant to compute Gower distance between two vectors.Similarly the module contains
tanimoto_gower
,tanimoto_gower_old
andtanimoto_gower_no_bool_weights
.My proposal is to accurately check the appropriate algorithm that can compute tanimoto-gower distance in an efficient way, and verify its results with tests in
test_metrics.py
.HINT:
tanimoto_gower
andgower_dist_numerical
do not seem to be working perfectly, but according to previous comparisons, they seem to lead to better results.The text was updated successfully, but these errors were encountered: