Open francoispichard opened 2 years ago
Hello @francoispichard,
Thank you very much for your kind words! We're glad that you enjoyed stream-learn
!
You're right, the variable self.current_ctdcs_
in fact contained a typo. This has now been corrected. Soon, we plan to review the library and introduce further improvements to adjust stream-learn
to the new versions of the required packages.
Thank you very much for highlighting this error. If you notice anything else, feel free to contact us.
Regards Pawel Zyblewski
Hello,
First of all I would like to congratulate you for developing this very nice package!
I have a question regarding the files
strlearn/ensembles/OOB.py
andstrlearn/ensembles/UOB.py
.If we look at line 28 in the OOB file and line 29 in the UOB file, we observe that the variable
self.current_ctdcs_
is declared. I wonder whether there is a typo in the name of that variable because it is never used. If I am not mistaken, the variableself.current_ctdcs_
should be replaced withself.current_tdcs_
because the former corresponds to the latter with an extra (unintended) character. This way, when considering the first instance in a new chunk (different from the first one), we can reuse the class sizes defined for the last instance in the previous chunk (for this we would needself.current_tdcs
=self.last_instance_sizes
), which would enable us to be consistent with equation (1) from ref. [1] (i.e., research article in which OOB and UOB were introduced) at all time points.Do you (@xehivs) also agree that this is a typo? Many thanks for your attention!
References: [1] S. Wang, L. L. Minku, and X. Yao, “Resampling-Based Ensemble Methods for Online Class Imbalance Learning,” IEEE Transactions on Knowledge and Data Engineering, vol. 27, no. 5, pp. 1356–1368, May 2015, doi: 10.1109/TKDE.2014.2345380.