Open kfuku52 opened 3 years ago
@Hego-CCTB I will take care of it if you don't have time.
I'll make a performance comparison, this shouldn't take too long. In terms of computing time, --one_outlier_per_iter to 'no' should always be at least equal or faster than 'yes', as there are fewer iterations involved. For actual quality, we'll need to see the amalgkit results. I'll run a couple of datasets in both modes and post the statistics here.
Have you had a chance to investigate this issue?
https://github.com/kfuku52/amalgkit/commit/12bf987bb7d022c4a86cc8693274086b80fb8e98 @Hego-CCTB This commit changes the default of --one_outlier_per_iter to 'no', to maintain the original behavior. We can activate it by default if its performance is consistently better, but I think we don't have any performance comparison yet.