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
mira_b
Level I

Showing Quantile Statistics Associated with Boxplots

Hi

I just started using JMP. When I create a boxplot through the graph builder it does not show the quantiles as shown in this example: https://www.jmp.com/content/dam/jmp/documents/en/academic/learning-library/02-box-plots-se.pdf

It only shows the boxplot. Where do I find the quantiles?

Thank you.

1 ACCEPTED SOLUTION

Accepted Solutions
Phil_Kay
Staff

Re: graph builder: boxplot, how to show quantiles

If you want the values for the quantiles you will have to use a different platform in JMP. Graph Builder is great for visual analysis. For statistics look in the Analyze menu.

Again, I would refer you to the one-page guide that you mentioned in your post. This shows you how to get the quantiles using Analyze > Distribution. If you want quantiles for the variable for different levels of some other variable(s) you can use the "By" role in the launch dialogue or the local data filter.

Alternatively you could try Tables > Summary.

View solution in original post

6 REPLIES 6
Phil_Kay
Staff

Re: graph builder: boxplot, how to show quantiles

Hi,

As the one-page guide shows, those box plots were created with other platforms in JMP: Analyze > Distribution, Analyze > Fit Y by X.

You can create box plots with quantiles in Graph Builder by changing the box plot type (lower left in the Graph Builder window).

I hope this helps.

Regards,

Phil

GB box plot with quantiles.jpg

mira_b
Level I

Re: graph builder: boxplot, how to show quantiles

Thank you for your answer, Phil.

I dont have problems creating  a boxplot (see my boxplot attached). I dont know how to show the quantiles (the acutal numbers, I mean).  I dont get a box with numbers. Where do I find this?

Thank you!

 

 

 

 

Phil_Kay
Staff

Re: graph builder: boxplot, how to show quantiles

If you want the values for the quantiles you will have to use a different platform in JMP. Graph Builder is great for visual analysis. For statistics look in the Analyze menu.

Again, I would refer you to the one-page guide that you mentioned in your post. This shows you how to get the quantiles using Analyze > Distribution. If you want quantiles for the variable for different levels of some other variable(s) you can use the "By" role in the launch dialogue or the local data filter.

Alternatively you could try Tables > Summary.

Re: Showing Quantile Statistics Associated with Boxplots

I would say Distribution and then show Quantile Box Plots. The table comes from Distrubiton platform

Chris Kirchberg, M.S.2
Data Scientist, Life Sciences - Global Technical Enablement
JMP Statistical Discovery, LLC. - Denver, CO
Tel: +1-919-531-9927 ▪ Mobile: +1-303-378-7419 ▪ E-mail: chris.kirchberg@jmp.com
www.jmp.com
KurtAnderson
Level III

Re: Showing Quantile Statistics Associated with Boxplots

I think "5 Number Summary" does the trick. In graph builder after creating a box plot, right click on the graph area, choose Box Plot, then "5 Number Summary". Max, Q3, Med, Q1 and Min are then displayed on the graph.

I wish the "5 Number Summary" were actually a "6 Number Summary". I would include N, the number of data points. When presenting box plot data, this is the most common question I get, "how many data points are represented by that box?" It's a more common question than the p-value or anything else.

Re: Showing Quantile Statistics Associated with Boxplots

The '5 number summary' was originally developed by John Tukey as part of his regular data analysis in the days before computer graphics. He later developed the 'box and whiskers plot' as the graphical representation of the same summary. So JMP is merely following the historical practice that is used by many analysts.

 

Your situation might be addressed by adding a plot of N for each group. That result is easily accomplished with Graph Builder:

 

Screen Shot 2020-10-30 at 5.35.14 AM.png

 

Or perhaps include the markers for the data to get a sense of the group sizes.

 

Screen Shot 2020-10-30 at 5.40.23 AM.png

 

The best solution depends on the particular details of your data, of course.