Timeline for Regression for power law
Current License: CC BY-SA 3.0
8 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Apr 28 at 22:24 | comment | added | Glen_b | A comment about running times on real machines; often the bigger problems will see a substantial change in what would otherwise be close to a power law because you can start seeing a lot of artifacts from the impact on memory usage -- swapping stuff that was being kept in memory out to disk and so on, which can slow things down. Meanwhile at lower size problems often the asymptotic behavior you want to pick up is impacted by the "less slow" parts (e.g if there's an $O(n^2)$ component and an $O(n\text{ lg } n)$ component with a large constant that second term might have a substantive impact) | |
| Oct 30, 2016 at 14:36 | history | edited | Glen_b | CC BY-SA 3.0 |
added 76 characters in body
|
| Oct 30, 2016 at 13:40 | comment | added | Glen_b | @kjetilbhalvorsen There's better ways to achieve that, but the fact that the model doesn't fit is a clear warning that if interest was to predict even slightly outside the range of the data, the predictions will be too low, perhaps dramatically so. The fit to only the last 21 points still doesn't fit the curvature within that part of the data. | |
| Oct 30, 2016 at 12:07 | comment | added | kjetil b halvorsen♦ | In this case, a fit that is better for the largest $y$-values might be appropriate, as we dont matter so much the running time when it is low ... | |
| Oct 30, 2016 at 9:07 | history | edited | Glen_b | CC BY-SA 3.0 |
added 296 characters in body
|
| Oct 30, 2016 at 8:54 | history | edited | Glen_b | CC BY-SA 3.0 |
added 28 characters in body
|
| Oct 30, 2016 at 8:42 | history | edited | Glen_b | CC BY-SA 3.0 |
added 272 characters in body
|
| Oct 30, 2016 at 8:31 | history | answered | Glen_b | CC BY-SA 3.0 |