Multiple Y-Axis with a Scaled Response Overlay Graph

[Steve_Kim]

Previously I asked a multiple Y-axis overlay chart

in the user community (Multiple Y axis in a chart. (Multiple Y axis scales))

and Byron made a scirpt for it (Multiple Y-Axis Graphs).

I hope I can get this feature from the 'Graph Builder' in next JMP version with interactive Y-axis setting.

Currently the multiple Y-axis graph (from the script) dosen't link with the data in the graph.

If the data is linked with multiple Y-axis interactively, it will be better than other supporting programs.

Thank you. : )

[atrautmann]

Dear James,

I just planned to create a similar wish but then I stumbled across your post.

You have my full support, a built-in function for the generation of multiple y-axis in graph builder would be extremely helpful.

Extended formatting features, e.g. colored y-axis, could improve the power of graph builder even more:

[Steve_Kim]

As you can see the following example which I found from the internet (I cannot post my company examples due to confidential reason),
'OSIsoft PI process Book' program allows a multiple Y-axis graph.
This type of graph is commonly used for biologics drug manufacturing companies, especially for the cell culture process.
We are still waiting for this new feature in JMP. Many thanks!

[Sarah-Sylvestre]

Hi @Steve_Kim, thank you for your suggestion! We have captured your request and will take it under consideration.

[Sarah-Sylvestre]

@Steve_Kim @atrautmann We are currently reviewing this request and I have a few questions for you both.

1. In order to make these graphs, do you typically leave JMP and use other software or use a workaround in JMP?
2. Would the following solutions satisfy this request. If not, why?
   1. Put each variable in its own lane on the Y-Axis (separate y variables against common x), paired with an improved crosshair tool that allows you to see the Y values at certain times/X values for each variable 
   2. Standardizing the y-variables and then overlaying them on the same graph. In this case, you obviously would not be able to see the original Y values but the standardized values, but it would show you how the trends of each variable compare to each other. I'm wondering if you really just need to compare trends, or if you also need to know the specific y-values for each variable at all times 
3. Is this a very standard way to graph data in your industry? Would alternative graphs that get the same message across be accepted, or is it a requirement to have a Multiple Y-Axis graph? If so, why? 
4. Would you rather see separate Y-Axis, like the picture posted by @atrautmann, or would it be alright to see similar to what @Steve_Kim posted, with just the min and max of each y variable represented? (see example I attached)

Sorry for all the questions! We are just trying to consider other solutions because the Multiple Y-Axis graph is generally not considered a best practice in the data visualization community.  Let me know what you think!

 

[FN]

Same topic with a similar number of votes here:
https://community.jmp.com/t5/JMP-Wish-List/Plotting-several-trends-from-sensor-data-that-have-differ...

[atrautmann]

@Sarah-Sylvestre: Happy to hear that you are reviewing this request! Please find the answers to your post below:

1. Yes, I typically leave JMP to make these kind of graphs.

2. Solution # 2.1 would be okay (having each variable on its own lane), this should do the job if the no. of variables is not too big. Solution # 2.2 is not satisfying, see the comment for # 3 below.

3. Yes, it is a common way to have multiply Y-Axis graphs in my industry (biotechnology/pharma). The reason is simple - imagine you want to trend a mixing process. You want to trend different variables that also differ in their min/max range, such as agitation (e.g. 0-150 rpm), pH (e.g. 6.0-8.0), Temperature (e.g. 20-40°C), oxygen content (e.g. 0-100 %), over process time. You need to see all these parameters at once and in one graph to have a full picture of the process and for a better process understanding. 

4. For me, it would be better to have separate Y-Axis as shown in my picture posted earlier. Or alternatively, the solution mentioned in #2.2, each variable on its own lane. This would allow a higher resolution of the single variables.

[Sarah-Sylvestre]

@atrautmann Thank you for the information! As noted by @FN, this request is the same topic as this one: https://community.jmp.com/t5/JMP-Wish-List/Plotting-several-trends-from-sensor-data-that-have-differ...

If it is alright with you all, I will mark this request as a duplicate and post all updates related to this functionality on the linked request. If you visit the other request, you will see that development has started working on this functionality but we aren't sure what version it will be available in. Let me know if you have any questions!

[atrautmann]

Fine for me, thank you Sarah