Timeline for Optimal way to creating new features from training set (in R)
Current License: CC BY-SA 3.0
9 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Nov 11, 2013 at 12:32 | answer | added | January | timeline score: 1 | |
| Nov 10, 2013 at 18:39 | comment | added | Rouse | I've updated the question with the setting where I am using. I am considering some greedy feature selection for feature selection. Are there any R packages that does this? | |
| Nov 10, 2013 at 18:35 | history | edited | Rouse | CC BY-SA 3.0 |
added 1101 characters in body
|
| Nov 10, 2013 at 13:07 | comment | added | cbeleites | Rouse, welcome to cross validated. In order to give you statistically sound advise, we need more information about your data, application and the model you want to use. Please explain what kind of measurements you data comes from, what application problem you want to solve, and what kind of models you consider. | |
| Nov 10, 2013 at 12:57 | answer | added | cbeleites | timeline score: 4 | |
| Nov 10, 2013 at 11:51 | comment | added | chl |
At this point, I'm afraid I'm not sure if this is not purely an R issue: there's a formula interface which allows to select all predictors and their interactions, e.g., y ~ .^2 will consider all variables and their first-order interaction. It then depends on what model you want to use to perform feature selection. Formulae are not always efficient in R, but loops aren't either. If your question is specifically about feature selection with such irregular data set, please clarify. Otherwise, it will be migrated to Stack Overflow (no need to cross-post).
|
|
| Nov 10, 2013 at 11:44 | history | edited | chl | CC BY-SA 3.0 |
added 2 characters in body
|
| Nov 10, 2013 at 11:39 | review | First posts | |||
| Nov 10, 2013 at 11:52 | |||||
| Nov 10, 2013 at 11:24 | history | asked | Rouse | CC BY-SA 3.0 |