Replies: 1 comment 2 replies
-
Not sharing the entire list is conceptual. Given the large number of oversamplers implemented, it is certain that there are bugs, misunderstandings, etc., and there were many more back when this study was created. Stating that an oversampler is the best (given our implementation) was found to be less harmful than stating that some oversampler is the worst among all (because this might be due to the misunderstanding of the algorithm, a typo in the paper from which we implemented, etc.). Therefore, intentionally, the lower end of the list is not disclosed. On the other hand, as numerous bugfixes were released in the past months, possibly a completely new evaluation should be executed. Nevertheless, the ones on the top of the ranking are certainly reliable ones, however, there is a chance that some new ones would show up in the top. |
Beta Was this translation helpful? Give feedback.
-
Currently requesting to see the longer rank list of the SMOTE competition as it seems interesting to see what other classification algorithms (with quadratic-like functions) can be injected e.g. RBF-based SVM, Naive Bayes, Gaussian Process, and QDA
Beta Was this translation helpful? Give feedback.
All reactions