Timeline for random forest variables importance with continuous and categorical variables and unbalanced output
Current License: CC BY-SA 3.0
5 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Nov 6, 2014 at 14:55 | comment | added | Donbeo | I think you should keep your answer maybe with a warning title or something like that. I think that these are the two main paper to consider. biomedcentral.com/1471-2105/9/307 biomedcentral.com/1471-2105/8/25 I had not the time to go deeply trough each paper but the problem and the solution seems to be pretty intuitive to me. | |
| Nov 5, 2014 at 13:19 | history | edited | rapaio | CC BY-SA 3.0 |
added 461 characters in body
|
| Nov 5, 2014 at 13:15 | comment | added | rapaio | First of all thank you for your link, I did not knew about it. It took me some time to parse it, but now I agree with the experimental setup and with the results. What do you think is better: delete may anser? or leave it like it is? I would not delete it for further reperence. PS: do you know about any further investigations regarding that? | |
| Oct 23, 2014 at 16:03 | comment | added | Donbeo | I do not agree with you. The importance computed with the gini index is biased respect to predictors with more unique values. This has been stated here biomedcentral.com/1471-2105/8/25 | |
| Oct 23, 2014 at 7:43 | history | answered | rapaio | CC BY-SA 3.0 |