Timeline for Encoding 'threshold-like' features
Current License: CC BY-SA 3.0
4 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Mar 30, 2017 at 12:32 | vote | accept | rinspy | ||
| Mar 29, 2017 at 9:29 | comment | added | rinspy | Also, I was assuming this would be a standard technique for such "threshold" data, but I was not able to find anything on the topic (data augmentation mostly brings up techniques on transforming image data). Do you know if there is anywhere I could find a discussion of transformations for the kind of "threshold" values I am dealing with? | |
| Mar 29, 2017 at 9:22 | comment | added | rinspy | Good point about augmented data being further from the boundary. I was just wondering if some non-linear models may 'assume' that output in some areas of the feature space may be anything, while I know what those areas of the feature space should output from the observations and knowledge about the domain. For example, I think KNN could benefit from having those additional data points? | |
| Mar 29, 2017 at 8:57 | history | answered | Hugh Perkins | CC BY-SA 3.0 |