Timeline for Is dimensionality reduction almost always useful for classification?
Current License: CC BY-SA 3.0
8 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Apr 17, 2015 at 2:03 | vote | accept | Tom | ||
| Mar 21, 2015 at 12:31 | answer | added | cbeleites | timeline score: 5 | |
| Mar 21, 2015 at 11:22 | answer | added | usεr11852 | timeline score: 2 | |
| Mar 20, 2015 at 10:17 | comment | added | amoeba | With dimensionality reduction of 20000 to 300 features, I will be very surprised if there is an actual real-life example where this would be detrimental for classification/prediction accuracy (even though theoretically possible). On the other hand, in many cases it will serve as regularization and will help. So I think the answer, in the context of your question, is Yes. | |
| Mar 20, 2015 at 10:05 | history | edited | amoeba | CC BY-SA 3.0 |
light editing
|
| Mar 20, 2015 at 6:51 | comment | added | ttnphns | Please read also this recent question: stats.stackexchange.com/q/141864/3277. It is not specifically about classification but is about prediction in general. | |
| Mar 20, 2015 at 6:12 | comment | added | usεr11852 | Your original features have a physical meaning, even something elementary. Your new eigenvectors, might have some meaning or might not. Therefore in the sense of interpreting your analysis reducing your number of features might not be always beneficial. Additionally there might be an issue that the first PCs are not good discriminants. With 300 PCs probably you are OK, but in smaller datasets one might omit relevant variation modes. | |
| Mar 20, 2015 at 2:42 | history | asked | Tom | CC BY-SA 3.0 |