Timeline for E(X1 | X2 > X3) for (X1,X2,X3) multivariate normal
Current License: CC BY-SA 4.0
10 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Jun 6 at 17:13 | history | edited | User1865345 | CC BY-SA 4.0 |
added 17 characters in body
|
| Feb 17 at 16:44 | vote | accept | frelk | ||
| Feb 17 at 16:13 | answer | added | ADAM | timeline score: 0 | |
| Feb 17 at 15:21 | answer | added | frelk | timeline score: 7 | |
| Feb 14 at 14:19 | answer | added | JimB | timeline score: 0 | |
| Feb 14 at 13:18 | comment | added | frelk | Thanks, that's very helpful. I think I could relax "closed form" to "easily evaluated with a computer." For example, if I need to do MCMC for an hour, that is probably too much. But if I all I need to do is evaluate the normal CDF or maybe even a numerical integral of a relatively simple expression that contains a term with the normal CDF, that is ok. Thanks! | |
| Feb 14 at 4:55 | comment | added | JimB | To echo what @MattF. mentioned, there won't be nice closed-form expressions mainly because of the need to integrate a normal CDF. The special cases for where there is a closed-form are likely not very useful. For example if all of the means are $\mu$ and variances are $\sigma^2$ and $X_1$ and $X_2$ are both independent of $X_3$, then the conditional expectation is $\mu+\rho_{12} \sigma/\sqrt{\pi}$. If $X_2$ is independent of both $X_1$ and $X_3$ (with as before all means and variances are equal), the conditional mean is $\mu-\rho_{13} \sigma/\sqrt{\pi}$. There might be other special cases. | |
| S Feb 14 at 3:35 | history | suggested | Preston Botter | CC BY-SA 4.0 |
formatting
|
| Feb 14 at 3:00 | review | Suggested edits | |||
| S Feb 14 at 3:35 | |||||
| Feb 14 at 2:54 | history | asked | frelk | CC BY-SA 4.0 |