Timeline for Plotting confidence interval from P values (from DistributionFitTest) for 2 parameters
Current License: CC BY-SA 3.0
6 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Oct 6, 2017 at 20:53 | vote | accept | Matt Stein | ||
| Oct 6, 2017 at 20:53 | comment | added | Matt Stein | I am 100% OK abandoning the P-value approach, and you're right, I'm just trying to get good estimations. I will refer to your other link for more help with the contours. Thank you again!! | |
| Oct 6, 2017 at 20:49 | comment | added | JimB | If you can bear with me a while longer: Forget the p-value approach, too. Your problem as stated involves "estimation" rather than "hypothesis testing". | |
| Oct 6, 2017 at 20:46 | comment | added | JimB | Forget 1/2/3-sigma: at best that applies to the one dimensional case. One is assuming/hoping that a bivariate normal distribution will closely approximate the bivariate distribution of the estimators of the scale and shift parameters. Search this site for "bivariate normal contours" or see mathematica.stackexchange.com/questions/153511/…. | |
| Oct 6, 2017 at 20:28 | comment | added | Matt Stein | Awesome, that's quick and easy. Now... plotting this... In theory, I should be able to position the "peak" at (momShift,momScale) on a 2D plot of the parameter space, and see rings go out at 1/2/3-sigma, but the exact shape/rotation of such a contour plot eludes me. For example, is it rotated at all? | |
| Oct 6, 2017 at 18:46 | history | answered | JimB | CC BY-SA 3.0 |