Warum ist die Fraction of Design Space Plot (FDS) Y-Achse in Varianz skaliert (DoE)

[#1]

Hallo,

ich würde gerne die y-Achse des FDS in Std. Abweichungen skalieren. Gibt es eine Möglichkeit oder einen Workaround?

Vielleicht kann mir ja auch jemand erklären warum die y-Achse auf Varianz basiert. Ich bin von anderer Software den "Std Error Mean" im FDS gewohnt. Was ist der Vorteil von Varianz gegenüber Std Abweichung?

Ich würde eine interaktive Antwort bevorzugen. Mit einen Skript kann ich auch umgehen.

Vielen Dank

Peter

[#2]

The Fraction of Design Plot shows the integrated prediction variance from the minimum prediction variance to the maximum prediction variance as larger fractions are integrated. It is actually the relative variance because the plot does not have information about the response variance. If you know the response variance, then multiply the variance in the plot by it to get actual prediction variance.

[#3]

Hello Mark, I switched the auto translate off. May be that corrupted my question.
I'd like to know, why is JMP in the FDS Plot using Variance. A squared term. Why is JMP not using std. error mean. A value I can easy work with. I would like to know why, then I have the chance to see the value in it, or I'd like to get the diagram with a transformed y-Axis (where is the formula for the trace or the look up datatable). That is what I'm used to work with from other Software.
Many thanks
Peter

[#4]

I honestly do not know why JMP chose to plot the variance instead of the standard error. The function from regression is a 'quadratic form.' They did not square the function. They chose not to transform it for the plot.

I don't think that the choice of standard error or variance makes any practical difference when using this plot to visually evaluate the performance of a design or to compare more than one candidate design. But remember that variance is additive, standard error is not. That property might give the use of variance an advantage.

JMP does not copy other software. JMP follows standards, but I am unaware of any standard for the FDS plot. Perhaps plotting the standard error is a convention in your experience, but I do not think that there is any convention either.

So what are you unable to do with the version of the FDS plot in JMP that uses variance?

You might suggest that JMP offers the option to plot either the variance or the standard error in the FDS Plot and Prediction Variance Profiler (Prediction SE Profiler, I guess) over in the JMP Wish List area of the JMP Community.