Timeline for Boosting AND Bagging Trees (XGBoost, LightGBM)
Current License: CC BY-SA 4.0
15 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Nov 19, 2018 at 21:49 | vote | accept | Jonathan | ||
| S Oct 29, 2018 at 0:02 | history | bounty ended | mkt | ||
| S Oct 29, 2018 at 0:02 | history | notice removed | mkt | ||
| Oct 26, 2018 at 21:09 | comment | added | EngrStudent | the bagged boosted is going to make a forest of series ensembles, and take the average output. It might engage the over-fitting that a series(boosted) ensemble can make, and give a more robust output, but the gain isn't going to be huge. | |
| Oct 26, 2018 at 21:08 | comment | added | EngrStudent | Here is the boosted-bagged. Instead of a new tree for each series-step you get a new forest with average output. Eugene Tuv and Kari Torkkola. jmlr.org/papers/volume10/tuv09a/tuv09a.pdf | |
| Oct 26, 2018 at 21:05 | comment | added | EngrStudent | you need a "compute the error" for your boosting. Done wrong that falls apart. The weights are critical for adaboost. It isn't a raw residual. ... We aren't talking about stochastic gradient as necessary in boosting, though it speeds things up. | |
| Oct 21, 2018 at 13:11 | answer | added | Laksan Nathan | timeline score: 22 | |
| S Oct 21, 2018 at 7:35 | history | bounty started | mkt | ||
| S Oct 21, 2018 at 7:35 | history | notice added | mkt | Draw attention | |
| Oct 19, 2018 at 13:57 | history | edited | Jonathan | CC BY-SA 4.0 |
[Edit removed during grace period]
|
| Oct 19, 2018 at 12:00 | history | tweeted | twitter.com/StackStats/status/1053254679864528896 | ||
| Oct 19, 2018 at 9:22 | comment | added | mkt | +1 for an interesting and very well-formulated question. And welcome to the site. | |
| Oct 19, 2018 at 9:06 | history | edited | Ferdi |
GBM is more specific than "Machine Learning"
|
|
| Oct 18, 2018 at 21:20 | review | First posts | |||
| Oct 19, 2018 at 0:27 | |||||
| Oct 18, 2018 at 21:15 | history | asked | Jonathan | CC BY-SA 4.0 |