Waterfall plots for signals not business in JMP?<meta http-equiv="Content-Type" content="text/html; charset=UTF-8" />Hi,
a customer asked if JMP can create plots usually used in Signal Processing, like a waterfall plot as this:
You have multiple measurements (Samples) for different X-Vars (C(X kHz)). As it could have Kind of dirac steps it is a very stiff Problem the Surface Plot cannot handle very well. Also with the Scatterplot 3D I could not come to a equivalent representation. May I have to Play with shading, light, ....but I believe the Problem is something different. We would not be able to create such dense steps in the steep Areas like a stiff ODE solver would create.
Of course I could send the data to MATLAB and create this waterfall plot. The question is can I do this also with JMP alone?
Thanks for your thoughts,
Martin
Re: Waterfall plots for signals not business in JMP?Hi Xan and Michael,
thanks a lot, you got it! I found my mistake: I used fc(kHz) instead of frequency and therefore did get a bad result. Now I get the same result.
The contour plots as well the scatterplot are good ideas for another view on the data which, as you said, could show patterns might not recognized in a 3D view.
Thanks a lot!
Martin
PS: Unfortunately I can only give one right answer, so I picked the first one in time and set Michael's as helpful at least.
Re: Waterfall plots for signals not business in JMP?These two are how I might represent this type of data... the surface plot is about what you are after, I think. modified table with scripts is attached.... might need a little tweaking of the color scheme to match what you are showing exactly.
Best,
M
Re: Waterfall plots for signals not business in JMP?What is the issue with Surface Plot? Too noisy? Too square? This is what I get after using a numeric rank for frequency to replicate the previous categorical treatment. (It's a question for the domain expert whether the frequencies should be spaced evenly or not.)
A scatter plot simulating heat map avoids some of the occluded feature issues.
I had to make it very wide, change the marker to a vertical line, and adjust the color scale to see the flare up near the bottom of the purple region. This does show a trough (green horizontal stripe) that I missed before. A true heat map would work about as well if you can accept the binning.
I can see it if I rotate the surface plot just right.
Re: Waterfall plots for signals not business in JMP?Hi Bill, I only found the one for business data which has a different purpose and meaning. If you have found a different one I'd be interested.
Re: Waterfall plots for signals not business in JMP?Hi Martin,
There is an add-in for Waterfall charts in the file exchange.
Bill
Re: Waterfall plots for signals not business in JMP?Ah, OK. Usually in Protein Sequencing using LC-MS-MS, one generates similar plots, but the small peaks are important as well.
Chris Kirchberg
JMP Technical Enablement Engineer
Worldwide Technical Enablement Team
SAS Institute, Inc.
100 SAS Campus Drive | Building T | Cary, NC 27513
Tel: +1-919-531-9927<>
Mobile: +1-303-378-7419<>
E-mail: chris.kirchberg@jmp.com
Unsubscribe<>
Re: Waterfall plots for signals not business in JMP?Hi Chris,
thanks for your answer. In Signal Analysis often you look just at the Spikes as the rest will be treated as noise. so if there is a small Change won't be that important unless you are looking for a specific pattern. This is as far I know mainly due to the fact that small changes would be filtered out to clean the Signal from the noise.
Martin
Re: Waterfall plots for signals not business in JMP?Hi Malcolm,
the NN is another option but still does not fullfill the needs. Anyway, thanks for the suggestion I believe this Approach can work for other use cases quite well. Below are some further trys.
Martin
Re: Waterfall plots for signals not business in JMP?Hi Martin,
You might be able to get close by fitting a NN then plotting prediction formula on a surface plot as in this example (no harm from forcing an overfitted model in this case):
[Missing image]
Note: I had to stack the data to get right shape of data, unstacked version also attached.
Malcolm
Re: Waterfall plots for signals not business in JMP?Hi Martin,
I don't think this can be done in JMP without some scripting. I personally not a big fan of these plots because small signals can get lost behind the mountains
Chris Kirchberg
JMP Technical Enablement Engineer
Worldwide Technical Enablement Team
SAS Institute, Inc.
100 SAS Campus Drive | Building T | Cary, NC 27513
Tel: +1-919-531-9927
Mobile: +1-303-378-7419
E-mail: chris.kirchberg@jmp.com
Unsubscribe
