About markschahl

markschahl

So a colleague of mine made some test samples and wants to compare them to commercial production results. See attached output. He labeled the test sample rows. He wanted to compare where they are relative the box plot of the commercial results. The only way he could see on the labeled rows and not the other rows was to hide the other rows in Graph Builder. Better would be to compare the test samples in the Distribution platform. Other than the Normal Quantile plot, how can just the test sample points shown on the box plots or the histogram? Is there a better way than the 'hide-the-rows' trick? Thanks!

markschahl · ‎06-04-2020

Peter: Thanks! That did the trick. I owe you a drink of your choice. I am working on the creators of the data to input the data in flat format...

markschahl · ‎06-03-2020

We all run into some version of this: Excel table formatted to look nice, no thought given to analyzability. In the free-form text cell, it looks like this: m/d: text [alt+enter] m/d: text [alt+enter] m/d: text [alt+enter] m/d: text [alt+enter] ... m/d: text [alt+enter] When the data are exported to JMP, the cell gets collapsed into: m/d: text m/d: text m/d: text ... m/d: text. I want to parse that long string with n elements into a separate row for each element, not split into separate columns. n is variable. This would create a new datatable that copies the other columns and adds the parsed element. Instead of Text-to-Columns, this is Text-to-More-Rows. Thankfully #/# is used consistently, so I could count those and loop n times. Has anyone out there done something similar?

markschahl · ‎02-26-2020

So, is t-SNE/UMAP going to be added to a future version of JMP? It is a very powerful abnormal event detection technique with wide applicability across industry. I just read this paper. t-SNE performs very well on the Tennessee Eastman Process dataset: A new unsupervised data mining method based on the stacked autoencoder for chemical process fault diagnosis

markschahl · ‎02-12-2020

Thanks Jim! I've never been to View > Addins before.

markschahl · ‎02-12-2020

Brady: Would it be possible for you to post the .jsl file that created the add-in? I have users that want to modify the output. Thanks!

markschahl · ‎02-01-2020

My company is not likely unique in that it uses Tableau for business informatics (notice I didn't say analytics, that's why we use JMP Pro). We have many thousands of Tableau users and many more thousands of people that use the Tableau browser interface. Those two populations outnumber our JMP users 3:1. I've been involved in many cases where the data came from Tableau dashboards, was exported to Excel, then imported into JMP. The problem solving was then done in JMP on this static data set. For example, we are applying Process Screening to a lot of non-manufacturing data: it is great for finding the signal in the noise. Having JMP and JMP Live live connect to Tableau servers would be great feature to: simplify workflows and promote real analytics, not to mention increase the JMP user base. This wish can of course be extended to other BI systems. @stan_koprowski - we had this discussion at Discovery 2019

markschahl · ‎08-20-2019

Batista: Help > Tutorials has two Join and one Stack tutorials. There are a few videos on JMP's YouTube Channel:

markschahl · ‎08-08-2019

Luccas: I would recommend dropping SAL on Color instead of Overlay since SAL is a continuous variable. That way you can assign a meaningful continuous gradient color theme like cool to warm (blue to red). See example of 6 bara water density vs entropy colored by temperature. Another tip for when you have many categories to overlay: use the color theme Spectral. Why? Lots of colors (so you don't have discern between 5 shades of light green...) and it follows the rainbow. In primary school we all (English speakers) learned ROYGBIV. This is great for timestamps as overlay: purples are in the past and reds are now.

markschahl · ‎06-14-2019

Scott: Thanks! I've been asked more than a few times by our users how to make this type of chart in JMP. It's very useful for visualizing volume and revenue for given products. I've linked this post to our internal Yammer group.

markschahl · ‎04-04-2019

Xan: Looks like your idea has been extended by Kaiser Fung into the Bar Density Plot. The bars are easy to interpret. The packed polygons allow display of areas that would be a sliver on a packed bar chart. Don't know if I am sold on the bar density plot yet... p.s. I used packed bars to analyze and present the secret stuff from above last month. Worked much better than a standard pareto!

markschahl · ‎03-29-2019

Phil - there seems to be some server problems when trying to downloading the .pdf and .docx files. No problem with the addin or the .zip. Using Chrome, the error returned is: community.jmp.com sent an invalid response ERR_RESPONSE_HEADERS_MULTIPLE_CONTENT_DISPOSITION

markschahl · ‎03-09-2019

I often go to NIST to get physical properties of fluids that I work with. This morning I needed properties of propene at 39 bara pressure from 70-80C (my day job involves propene to make poly(propene), aka polypropylene - plastic #5). The NIST webpage returns a graph and options for viewing as a table or dowload as tab-delimited text file. Pretty useful. The joy comes when you: 1) Copy the URL: https://webbook.nist.gov/cgi/fluid.cgi?Action=Data&Wide=on&ID=C115071&Type=IsoBar&Digits=5&P=39&THigh=80&TLow=70&TInc=1&RefState=DEF&TUnit=C&PUnit=bar&DUnit=kg%2Fm3&HUnit=kJ%2Fkg&WUnit=m%2Fs&VisUnit=Pa*s&STUnit=N%2Fm 2) in JMP: File > Internet Open; paste URL; Open As = Data There you have it, a new JMP data table with thermophysical data. I will be regressing density(temperature) and other properties for use in a model that I am developing. If you have not used the NIST Chemistry Webbook, test it out.

markschahl · ‎02-14-2019

Just found this and used it. Thanks! One thing I found is that the list can truly take relative sizes not in pixels. I built a two panel graph with [125 375]. I tried [1 3] just for grins. Got the same result. DO NOT comma separate the elements in that list. It won't run. Took me 20 minutes to figure out what my mistake was...

markschahl · ‎01-29-2019

Thanks! From your script I know see why mine did not work: Parse( "\!"" || Char(ValuesList[i]) || "\!" = \!"" || LabelsList[i] || "\!"" ) )

markschahl · ‎01-29-2019

Jim: Thanks for the elegant solution!

markschahl · ‎01-29-2019

Hi! I have two x-variables: Meeting.# is numeric ordinal (just the Row()) Meeting.ID is character nominal (can be date or some other text descriptor) I need to plot attendance (X_2-X_8) against Meeting.# so that I can get a trend. But I want to display Meeting.ID as the x-axis labels. I tried this approach: https://community.jmp.com/t5/Discussions/How-to-Add-Value-Labels-to-a-Column-through-JSL/td-p/41604 But had issues with type-mismatch. There are other applications of this problem: Experiment.# and Experiment.Description Is there a better way to do this? Should the be done with Column Property Value Label or as part of Graph Builder? Ditto Value Ordering Should I submit a feature request? Thanks!

markschahl · ‎12-14-2018

Visualizations superior to pie chart: Analyze > Quality & Process > Pareto Plot (automatically sorted) Packed Bars: https://community.jmp.com/t5/JMP-Blog/Introducing-packed-bars-a-new-chart-form/ba-p/39972 #onelesspie

markschahl · ‎11-30-2018

Thanks! That suggested work-around did the trick.

markschahl · ‎11-15-2018

As a recent upgrader to JMP Pro, I now have access to the Reliability Simulation tools. Previously, I had used RAPTOR from the USAF to do this type of work. A common problem in logistics at a refinery or integrated chemical complex is sizing of buffer vessels between sequential units. Example: Producer-A -> buffer tank -> User-B I know: User-B time to failure distribution (TTF) - a nice Weibull fit User-B time to repair distribution (TTR) - a nice logNormal fit In the options for Time to Completion there is no option logNormal distribution, only Normal. How long User-B is down is the information needed to calculate the size of the buffer tank so that Producer-A can avoid shutting down while User-B is down. Two questions: 1) Is there good way to simulate logNormal distributed Time to Completion with JMP Pro 14.1? 2) Any plans to add other distributions like logNormal to this platform in the future (or should I add to the wish list)? Thanks!

markschahl · ‎11-06-2018

As Mark states, there is no substitute for a designed experiment (and some domain knowledge). Correlation does not necessarily imply causation. Case in point: http://tylervigen.com/spurious-correlations

markschahl · ‎09-11-2018

JMP Family: You now find yourselves in the same predicament I was in a year ago. Being able to follow a local river/creek gauge during Harvey was very helpful to understand what was going on. Here is a link to one in Raleigh. You should be able to find others close to your homes: Flood Gauge Raleigh, NC Good luck. Stay dry. See you safe and unscathed @Discovery!

markschahl · ‎08-27-2018

Yes, they exist in standard JMP. Links to a video and blog post below: https://www.youtube.com/watch?v=9KZ7Ns3CQzU https://community.jmp.com/t5/JMP-Blog/Scenario-based-planning-amp-optimization-in-Excel-amp-JMP-Part-4/ba-p/30406 Dear JMP Developers / Systems Engineers: Making a video of the blog post would be a great addition to the YouTube channel...

markschahl · ‎07-24-2018

See screenshot. JSL: If( Date Difference( :A, :B, "day" ), 1, 0)

markschahl · ‎07-24-2018

Don't be afraid, Formula Editor is your friend! Take it from someone that presented on it a previous Discovery Summit: Discovery-Summit-2016: Letting-Go-of-Excel-Hello-Formula-Editor As far as checking your work goes: make a distribution of your new column. Better practice for the plots is to also plot the input variables so that you verify the logic is correct. JMP 14 lets you preview your results at the bottom of the Formula Editor window (nice addition, developers!). In this case, the formula is pretty simple. I analyze a lot of time series process data (thousands of rows of pressure, temperature, flow, etc.) and have to write formulae to find conditions (P > x AND T > y). In those cases, I will also plot the inputs and formula result versus timestamp to see if the formula does what I want. Here's a Mastering JMP Webinar on the subject by Jerry Cooper (my partner in the video above)

markschahl · ‎07-21-2018

Thanks, Jerry! Worked nicely. Many folks in the company will leverage both solutions to this post.

markschahl · ‎07-21-2018

Thanks! Worked perfectly. I am using JMP Pro 14.

markschahl · ‎07-20-2018

Hi! Got a request from one our users in refining to interpolate a crude distillation curve. Some background info: YouTube Video on Distillation Plots ASTM D7169 Data table has twelve columns with the boiling points for different volume%iles of the crude. I want a formula that will interpolate across the columns to calculate how much has a boiling point less than T.Of.Interest. I know that the Interpolate() function does this. I also know that I need some way to get the values of the %ile columns into a list, probably using Get As Matrix(). I have yet to venture into the Matrix... Interpolate( T.Of.Interest, [0.05, 5, 10, 20, 30, 40, 50, 60, 70, 80, 90, 95], [<the columns with the boiling points>] ); How do I get the column values into a list? Or, is there a better way to do this? Without transforming the data table, is there a way to plot the BP list versus the %ile list? Thanks!

markschahl · ‎07-20-2018

Jerry: Thanks! That worked! Applogies for the delay. Almost three months of travel, including training 100+ engineers at five sites globally how to use JMP, delayed my response. It's been just a bit busy...

markschahl · ‎07-19-2018

This is awesome! I use parallel plots a lot for process troubleshooting. This is a great enhancement. I owe you two drinks now 🙂

Date Registered	‎06-18-2012 04:36 PM
Date Last Visited	a month ago
Total Messages Posted	88
Total Kudos Received	50

Online Status	Offline
Date Last Visited	a month ago