Timeline for R's lm prediction interval vs simulation
Current License: CC BY-SA 3.0
7 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Apr 13, 2017 at 12:44 | history | edited | CommunityBot |
replaced http://stats.stackexchange.com/ with https://stats.stackexchange.com/
|
|
| Mar 3, 2016 at 8:57 | vote | accept | Alireza | ||
| Dec 6, 2015 at 0:15 | comment | added | atiretoo | At that sample size you need is confidence limits on your confidence limits to be able to say if they are different or not! | |
| Dec 6, 2015 at 0:08 | comment | added | Alireza | Thanks! It's much better now, but it still doesn't match it completely though, especially for smaller n, e.g. n = 5. | |
| Dec 6, 2015 at 0:02 | comment | added | atiretoo | The prediction interval from predict is using a t-value based on the sample size, which will approach 1.96 as it gets larger. Try replacing 1.96 in your simulation with qt(0.975,n) | |
| Dec 6, 2015 at 0:00 | comment | added | Alireza | As I decrease/increase the number of points, the gap between my simulation and predict.lm's result increases/decreases. I can't figure out exactly why though... What am I missing here? | |
| Dec 5, 2015 at 23:31 | history | answered | Alireza | CC BY-SA 3.0 |