Timeline for XGBoost - does it make sense that accuracy decreases as threshold increases?
Current License: CC BY-SA 4.0
4 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Aug 20, 2021 at 8:05 | comment | added | user2974951 | @albus_c Yes, in this specific case because you coded the classes as such. | |
| Aug 20, 2021 at 7:54 | comment | added | randomal |
To be sure I understood correctly: the logic is if probability > threshold, then minority classes. Else, majority class. Is that correct?
|
|
| Aug 20, 2021 at 7:53 | vote | accept | randomal | ||
| Aug 20, 2021 at 7:49 | history | answered | user2974951 | CC BY-SA 4.0 |