cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 
Check out the JMP® Marketplace featured Capability Explorer add-in
Choose Language Hide Translation Bar
jmpneeeewbie
Level II

How to show mean +/- 3SD range in graph builder?

I want to show and compare the values by showing its mean+/-3SD range in graph builder.

The graph builder only provide the bar range of mean+/-SD.

 

2 REPLIES 2
SDF1
Super User

Re: How to show mean +/- 3SD range in graph builder?

Hi @jmpneeeewbie ,

 

  Not sure if this is exactly what you want, but it is ONE way of getting what you want (I'm using the Big Class example table from JMP help):

SDF1_1-1675442143859.png

 

  To get this, go to graph builder, select the histogram graph option and cast a column to the Y role in graph builder. If you then double click the Y axis, you can edit the setting for the axis and add some reference lines. I got the +/-3SD values by performing a histogram analysis on :height and then just calculating the +/-3SD values based on the SD of :height. In the Reference Lines sub panel, you can enter in the values for the lines, change the line colors and even line styles. 

SDF1_2-1675442255106.png

 

  This is just one way it can be done.

 

Hope this helps!<

DS

Victor_G
Super User

Re: How to show mean +/- 3SD range in graph builder?

Hi @jmpneeeewbie,

 

For Control chart, is there a reason for using Graph Builder and not Control Chart Builder (in Analyze, Quality and Process, Control Chart Builder) ?

You just have to select your Y response, and in the left panel, check in the submenu "Limits[1]" the options "Zones" and "Shade Zones". Once I have checked these options and click on done, I have this control chart (for Big Class dataset) :

 

Victor_G_1-1675443833260.png

And you can easily spot the +/- 1 SD zone (C, in green), +/- 2 SD (B) and +/-3 SD (A) in yellow.

Maybe what not you expect, but I just wanted to make sure you know this option in Control Chart Builder.

Victor GUILLER
L'Oréal Data & Analytics

"It is not unusual for a well-designed experiment to analyze itself" (Box, Hunter and Hunter)