About Thierry_S

Thierry_S

A quick fix: hold "CTRL" "SHIFT" "-" to decrease the overall font size across all display in JMP

Thierry_S

Hi, One possible thing that could be causing the difference in chart size despite defining the size of the FrameBox is the length of labels in your Y or X axis. Is it the case for your application? Another possibility would be to call the FrameSize command after the graph has been generated: report (/your graph variable/) [FrameBox (1)] << Frame Size (600,400); Just a couple of thoughts. Best, TS

Thierry_S

Hi, Two low tech solutions: 1) Omit the Label variable in the transpose parameters. Limitation: the columns will be labelled by ROW numbers 2) Create a dummy label column by concatenating the Label and Column (:Label || "," || :Column) and use it as Label in the transpose parameters There is a chance that neither suggestions will meet your expectations, therefore let me know if there are any additional information you could share.

Thierry_S

Oops, txnelson provided a much nicer solution. Best, TS

Thierry_S

summaryABC = dt<< Summary(Invisible, Group (:A,:B,:C); // This hides the Summary ABC table SummaryAB = summaryABC << Summary(Group (:A,:B); // other misc table operations //turn on screen update? summaryABC << Bring Windows to Front; // This brings Summary ABC to the front Have you considered using the "Invisible" switch in your Summary command (see above)? You can then bring back the window with "Bring Window to Front". Note: I have not found the exact command line to turn on and off the hidden status of a window. Best, TS

Thierry_S

Hi, I'm not sure if I'm correct but it looks like the two-way clustering data was standardized by columns (standard option for hierarchical clustering) while the recreated heatmap is using the original scale. If it is the case, you can easily standardize your data by selecting the data columns > right-click on the header > select New Formula Column > Distributional > Standardized. Then, use the standardized column as your Color variable in your heatmap. I hope it helps. Best, TS

Thierry_S

Hi Mark, Thank you for your feedback: this is exactly why I asked the question. I'm indeed concerned about setting unreasonable expectations with this biased approach: when I present ROC analysis data using a representative patient population (random selection), I'm often told that the results from my cherry-picking colleagues are much "stronger". Hence the need for solid justification on my part. Best, TS

Thierry_S

Hi, Thank you for your feedback. Just to clarify, the example cited in my original post is about exploratory biomarker research and not about clinical development: decision making about patients' health is never conducted that way. Best, TS

Thierry_S

Hi, A couple of notes (Sorry, I don't have the time right now to provide a complete solution) 1) Specify the format of the output of the "Column Get Names" function as text by replacing the second line of your code with: ColNamesList = dt << Get Column Names(string); you won't need to use the statement "Char" function in your code below 2) When deleting an unknown number of columns, it is best to start from the last column and to progress backward: For (i = NCols, i>=1, i--, .... 3) You may want to simplify your delete line of code by simply using: dt << Delete Column (Column (i)) Best, TS

Thierry_S

You are welcome. The Label variable is used to annotate the dendrogram: it is not really necessary but it useful to figure out visually which rows are closely related. Of note: by clustering your participants, you will create subgroups that are internally correlated (basic principle behind clustering algorithms). Therefore, you will need to be careful with the interpretation of any correlation analysis you may choose to conduct in the sub-groups. Best, TS

Thierry_S

Hi JMP Community, When conducting a logistic analysis (i.e. Y = Categorical and X = Continuous), are there any assumptions in data structure, distribution, and/or sampling method especially in regard to ROC analysis? Specifically, in clinical research aimed at identifying predictors of response to treatment, I have noticed that some of my colleagues select patients with the best response and the worst response, excluding the "middle" of the response distribution. Upon identification of a candidate predictor, they go on to produce ROC analyses (same biased data set) where the AUCs appear to be quite inflated. Hence, does the intentional selection of patients violate the statistical assumptions built-in ROC analysis? Thank you for your help. Best, TS

Thierry_S

Hi FP, To the best of my knowledge, I think that you can achieve the desired outcome by using the Clustering > Hierarchical Clustering on the transposed data where Items are in columns and Participants are in rows. Goto Analyze > Clustering > Hierarchical Clustering Select the Items columns > click on Y, Columns Select the Participant column > click on Label Uncheck the "Standardize Data" Click OK On the clustering result window, you can select the number of clusters by sliding the diamond handle and then save the Clusters and/or the Display Order I'm aware that this might be way too simple for your need but it could get you started. Best, TS

Thierry_S

Hi, Here is a possible solution to your problem: the Set Max Selected needs to be applied to the colList variable separately: Names Default to Here (1); dt = Current Data Table (); nw = New Window("TEXT",colList = Filter Col Selector( dt, Width( 256 ), N Lines( 20 ))); colList << Set Max Selected (2);

Thierry_S

Hi, I'm not sure that I can help but would it be possible to provide some clarification / additional details about your question? 1) Could you confirm that the data table is composed of 15 items (in rows) with ranking between 1 to 15 provided by 25 participants (in columns)? 2) Do you have a criteria for defining homogeneity of ranking? Is it based on a reference ranking (e.g. by an expert), on an average ranking, or something else? 3) When you mention that there is no agreement in ranking across all 25 participants, what was the exact procedure you used to evaluate Kendall's Tau? With this, I or other may be able to help you. Best, TS

Thierry_S

Hi Rob, It might be obvious but based on the JMP help section, the command Filter Col Selector does not appear to have an option for specifying Maximum selected items. Best, TS

Thierry_S

Hi, This might be obvious but from the screenshot of your code, it is not clear that you are actually pointing to the Graph Builder object when you use the Save Picture command (i.e. your last line of code should be "obj << Save Picture (Path, Format);" where obj should be defined as the entire code define under "Graph Builder").

Thierry_S

Hi, You just need to help JMP figure out that the variable "use_cols" is referring to columns by adding "as columns" to your last statement: dt << New Column( "max_col", numeric, formula( Maximum( as columns( use_cols ) ) ) );

Thierry_S

Hi, This a functionality that most of us would love to have but it is not available, at least in JMP14.1. One of the most efficient way to add interactions between variables in existing model is to: 1) run the model as-is 2) save the script: red top left triangle menu > Save script > To Script Window 3) In the Effects line of code, add the interactions you which to explore (make sure to use the exact syntax for variables as listed for the individual effects) Example: :data1 * :data2, *:data1 * :data3....

Thierry_S

Hi, Have you look into the "Filter" option in the Fit Model dialog screen? In the "Select Columns" area, click on the red triangle and select "Show Filter". Search options are accessible via the black triangle just to the right of the Text Box (next to the magnifying glass icon). If this not what you are looking for, you may need to consider scripting which will require sharing your an example of your data (if possible). Best, TS

Thierry_S

Hi, Right click on the graph body and select Customize > Select the "Bar" item and increase the "Width Proportion". Note, you may need to bump that number to 100 or even larger values if you have a lot of data. Best, TS

Thierry_S

Hi JMP Community, I would like to propose an enhancement to the scripting of Graph Builder plots which would enable the addition of simple graphic elements (line, boxes, arrows...) to the Axis region (AxisBox) of the graph (the current GB scripts are limited to the GrahBox region). Here is an example of the type of enhancement I have in mind:

Thierry_S

Hi Jim, Thank you for your feedback: it always very appreciated. Should I assume that this specific feature does not exist yet and therefore it would be appropriate to suggest it on the Wish List? Best regards, TS

Thierry_S

One note: the recoded Pass Fail needs to be numerical but it can be nominal or ordinal too

Thierry_S

Hi Satej, The simplest solution is to recode your Pass_Fail column as a Continuous numerical column with Fail = 0 and Pass = 1. Then you can just look for the minimum by Sr ID to determine if any of the Sr ID failed at any time. See this solution attached Best, TS

Thierry_S

Hi JMP Community, Is it possible, via scripting, to dynamically highlight axis values (see image attached) with transparent color boxes without having to first convert the graph to an image and work with drawing / pixel commands? Thank you for your help. TS

Thierry_S

Hi, Have you considered using the option of Label by Row: this should hold the format define in your table

Thierry_S

Hi, I suspect that you may have built your Pie chart in an unusual manner. Indeed, on your screenshot, there is no variable associated with the Y axis. The Y Axis should contain the counts variable for calculating proportions in the Pie chart and X Axis contains the categories for each the slice of you pie chart. In order to confirm this, would it be possible to expand the last field under the left "Pie Menu" > "Variables" Best regards, TS

Thierry_S

Another quick (and dirty trick) is to hold "CTRL" + "SHIFT" + "+" to increase the overall font size across all elements (table, plots...) in JMP To reduce font size, hold "CTRL" + "SHIFT" + "-"

Thierry_S

Hi, Have you tried changing the patterns by right-clicking directly on the legend element you are trying to modify (not via the Legend Settings option in the main graph menu)? That's the only thing I can come up with. Sorry I cannot be of more help. Best, TS

Thierry_S

Hi again, Forget about my comment on the Legend placement inside the graph frame: I was not paying full attention when I wrote that. So, with the legend inside the graph frame, I'm still not able to reproduce your problem. Hence, understanding how you changed the bar patterns may provide insight on how to fix your problem.

Date Registered	‎04-19-2018 01:39 PM
Date Last Visited	3 hours ago
Total Messages Posted	180
Total Kudos Received	88

Online Status	Offline
Date Last Visited	3 hours ago