Timeline for Is there a variant of PCA that returns components wholly or not at all, ie, 0 or 1 coeff instead of ratios?
Current License: CC BY-SA 4.0
6 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Aug 25, 2023 at 21:16 | history | edited | Nick Cox | CC BY-SA 4.0 |
deleted 1 character in body
|
| Aug 25, 2023 at 17:57 | comment | added | Galen | I'm not sure how to do it with the usual approach based on singular value decomposition. But you could approach this with Bayesian canonical polyadic decomposition. | |
| Aug 25, 2023 at 17:54 | history | edited | kjetil b halvorsen♦ |
edited tags
|
|
| Dec 19, 2019 at 22:46 | comment | added | Jake Westfall | What you're describing reminds me of a technique called Interpolative Decomposition. Here's a useful twitter thread that walks through a short case study: twitter.com/seanjtaylor/status/1095567688657559552 | |
| Dec 19, 2019 at 22:14 | history | edited | Geoffrey Anderson | CC BY-SA 4.0 |
elaborated
|
| Dec 19, 2019 at 22:04 | history | asked | Geoffrey Anderson | CC BY-SA 4.0 |