Under Tables > Stack, the option for "multiple series stack" is  very helpful in some data situations. Would it be possible when "multiple series stack" is selected to have a preview box to show where the columns would go for the two options: (1) every nth column vs (2) n columns in a row (contiguous)? If Number of Series is used, preview panel box would show which columns would go in 2, 3, ... n groups, or highlight the first of the series in the upper column dialog. So if 3 was entered, the 1st, 4th, 7th, etc column selected would be highlighted above to safe space. If Contiguous is checked, would show which columns in how many groups of contiguous columns are being stacked.   From the JMP on Air topic: https://community.jmp.com/t5/JMP-On-Air/The-Tables-Menu/m-p/261014#U261014    ... View more

Just watched the JMP on Air segment on Statistics Index, The Statistics Index. I like to tell new JMP users about it in JMP classes.   Would like to suggest adding a filter like in the Scripting Index or Preferences. Right now in the Statistics Index you can start typing in the menu to find something, but it is a first character look-up. I search for Steel-Dwass and find it, but if I only remembered something like "Dwass", then it will not show. Or say you want to find any topic on "design", it is scattered throughout the index like "experimental design", "factorial design", "fractional factorial...", etc. ... View more

A low priority request, but could help new JMP users. The term "Kruskal-Wallis" only shows up when you hover over the term "Wilcoxon" in the Fit Y by X for nonparametric, or after you run the analysis. After running, it shows "Wilcoxon/ Kruskal-Wallis Tests (Rank Sums)".   In preferences, when you search the term "Kruskal", it does highlight the Oneway platform and the Wilcoxon term.   Other software shows "ANOVA" and "Kruskal-Wallis" side by side. When teaching the workforce which test to use, in other software can point to the Kruskal-Wallis as a menu item, but in JMP, have to remind them it is the Wilcoxon menu item, and can see it will be the Kruskal-Wallis test for more than 2 items.   Could the actual term "Kruskal-Wallis" be shown in the menu item? Such as "Wilcoxon/Kruskal-Wallis" as the menu item under the nonparametric tests?   This is similar to having "Wilcoxon Each Pair" vs "Steel-Dwass All Pairs" clearly shown in the nonparametric multiple comparisons. ... View more

  Is anyone else bothered by the orientation of the contingency table and mosaic plot? I do not use it, but one of our JMP users wanted to flip the orientation of the table so that it matches the mosaic plot. Found out that if you swap X and Y, both the table and mosaic plot update. See script below to run an example and compare them side-by-side. It seems you would want the numbers in the table to line up with the cells in the mosaic plot for easier viewing and referencing. I could not find any previous discussions. It appears the contingency table cannot be changed, it is a CrossTabBox, which does not lend itself to changing anything. The script below has in the top part the work around, which is to make a clone of the table made when switching the x and y, and then appending that to the mosaic plot. Also had to do a little with Value Ordering to get it to line up. Will add to the JMP wishlist. I did discuss at JMP Discovery Summit 2018 with the EA lab and the developers.     Found some posts on R language, where you can use the dir() argument and specify the horizontal and vertical, while the "count" output in RStudio stays the same for the orientation. # R Mosaic Plot Example - Changing Direction getwd() employee <- read.csv("Products.csv", TRUE, sep = ",",                      na.strings = TRUE) count <- table(employee$EnglishCountryRegionName,                employee$Color) count mosaicplot(count, main = "Countries Mosaic Plot",            sub = "Product Colors by Country",            xlab = "Colors",            ylab = "Countries",            las = 1,            dir = c("h", "v"),            color = "skyblue2",            border = "chocolate") // For a contingency table report, swap the x and y // currently can swap the mosaic plot only Names Default to Here(1); dt = Open( "$SAMPLE_DATA/Big Class.jmp" ); myr = Contingency( Y( :Age ), X( :sex ) ); // Have to have a Contingency Analysis report open // Can add test to make sure it is open in the future myr = Current Report(); // taking apart CrossTabBox was easier using XML than JSL mys = myr["Contingency Table"][CrossTabBox(1)] << Get XML; // get current X role or rows of contingency table start = Contains(mys,"<CrosstabBoxRowsLabel>")+22; end = Contains(mys,"</CrosstabBoxRowsLabel>"); newY = Substr(mys,start,end-start); // get current Y role or cols of contingency table start = Contains(mys,"<CrosstabBoxColsLabel>")+22; end = Contains(mys,"</CrosstabBoxColsLabel>"); newX = Substr(mys,start,end-start); // Recreate same contingency table but with X and Y flipped // also temporarily use Value Order to match the Mosaic Plot myValueLabels = Reverse(Associative Array(Column(dt,newX)) << Get Keys); Column(dt,newX) << Set Property("Value Ordering",myValueLabels); tempCont = Contingency( Y(Column(dt,newY) ), X(Column(dt,newX) ), Contingency Table ); tempContr = tempCont << Report; mycopy = tempContr["Contingency Table"][CrossTabBox(1)] << Clone Box; tempCont << Close Window; Column(dt,newX) << Delete Property("Value Ordering"); // remove the original contingency table and replace with transpose myr["Contingency Table"][CrossTabBox(1)] << Delete; myr["Contingency Table"] << Append(mycopy); // show original side by side obj = Contingency( Y( :Age ), X( :sex ) ); Report(obj)["Contingency Table"] << Set Title("Contingency Table - Current output"); myr["Contingency Table"] << Set Title("Contingency Table - Wish List for JMP16");   ... View more

Level: Intermediate   Daniel Sutton, Statistician - Innovation, Samsung Austin Semiconductor       The CUSUM Control Chart has been shown to be sensitive to small shifts compared to standard WECO (Western Electric Co) rules and Nelson rules, in particular Rule 2: “9 points in a row on a single side of the center line” [1]. Some examples of CUSUM control charts show the CUSUM technique will pick up a signal in a sequence of 9 or more points, where just one point is barely on the other side of the center line, and so Rule 2 is not triggered in that particular sequence. Question: is the SPC chart missing the signal because we do not allow for the one point on the other side of the center line? The alternative for Rule 2 of “n out of n+k points” is re-visited. Control charts in JMP and other software are hard-coded for Rule 2 “n points in a row on a single side of the center line”. This might require updating the Control Chart Builder options for Customize Tests to allow this configuration for the user, so that one platform, the Control Chart Builder, can continue to be used for all SPC analysis.      Utilizing JMP software’s built in reports of Process Screening, Control Chart Builder, and CUSUM, JMP scripts were created that could quickly generate random normal data and example signals, analyze combinations of varying k and n for the “n out of n+k” rule, and at the same time compare to traditional tests and CUSUM. The scripts extracted when tests were triggered, determined how many signals were detected, and how many false positives and false negatives occurred. Examples for different scenarios were compared.      The results show that CUSUM is better for both false positives and detects small shifts faster than even “n out of n+k” rules. Rules 1,2,5,6 do better for short duration signals. A drawback of the CUSUM chart is that it does not display the actual data points. It is recommended that the CUSUM chart be added to Control Chart Builder, as an option along side the standard SPC charts, for graphical interpretation in a single window for the user. If the CUSUM chart cannot be easily added to the Control Chart Builder, the alternative of adding a user configurable “n out of n+k” rule is also recommended for the Control Chart Builder platform. ... View more

For future Control Chart Builder enhancements in JMP15 or beyond. New drop zones for columns that contain multiple LSL, LCL, target, UCL, USL values.   Some users have to track multiple spec limits and control limits in a relational database, so columns are stored for LSL, LCL, target, UCL, USL. Would be nice to have drop zones on Control Chart Builder for these columns, instead of relying on the static Column Properties for Spec Limits and Control Limits. For example, in Graph Builder, we can drag these columns to the right axis, and change these to lines, while keeping the raw data on the left axis as points, which mimics a control chart with changing spec limits or control limits. Or if different limits, can use these drop zones and use the By button to plot by what changes, like process or product. In Control Chart Builder, can somewhat mimic by using the Phase drop zone for a column like target, if it changes, and then use a script to add the other lines.   These drop zones would have to override any Column Properties if there are conflicts, or make a warning. And you could not save these to the column properties, since they change vs. sequence. ... View more

For each version of JMP, having to turn on in preferences the VIF column for Parameter Estimates in Fit Least Squares. I teach the Regression class in JMP, and always have to show new users that you have to right-click over Parameter Estimates table in the report to select VIF from the columns that are not displayed by default. I then have to show the class how to turn this on in the Preferences. One suggestion would be to allow this to be shown with a drop down menu item. From the Least Squares Fit Inverted Red Triangle:    Under menu item Regression Reports, the Parameter Estimates can be turned on or off. (on by default)    But also under this menu, selecting the "Show All Confidence Limits" turns these two columns on in Parameter Estimates, and in Expanded Estimates (but not a column available in Sorted Estimates). So can "Show VIF" be added to this menu just like "Show All Confidence Limits"? It would be easier to find instead of hidden in a right-click on the table in the report. Saw this question a couple of times in JMP community as well. Wish list for JMP15. ... View more

// For a contingency table report, swap the x and y // currently can swap the mosaic plot only Names Default to Here(1); dt = Open( "$SAMPLE_DATA/Big Class.jmp" ); myr = Contingency( Y( :Age ), X( :sex ) ); // Have to have a Contingency Analysis report open // Can add test to make sure it is open in the future myr = Current Report(); // taking apart CrossTabBox was easier using XML than JSL mys = myr["Contingency Table"][CrossTabBox(1)] << Get XML; // get current X role or rows of contingency table start = Contains(mys,"<CrosstabBoxRowsLabel>")+22; end = Contains(mys,"</CrosstabBoxRowsLabel>"); newY = Substr(mys,start,end-start); // get current Y role or cols of contingency table start = Contains(mys,"<CrosstabBoxColsLabel>")+22; end = Contains(mys,"</CrosstabBoxColsLabel>"); newX = Substr(mys,start,end-start); // Recreate same contingency table but with X and Y flipped // also temporarily use Value Order to match the Mosaic Plot myValueLabels = Reverse(Associative Array(Column(dt,newX)) << Get Keys); Column(dt,newX) << Set Property("Value Ordering",myValueLabels); tempCont = Contingency( Y(Column(dt,newY) ), X(Column(dt,newX) ), Contingency Table ); tempContr = tempCont << Report; mycopy = tempContr["Contingency Table"][CrossTabBox(1)] << Clone Box; tempCont << Close Window; Column(dt,newX) << Delete Property("Value Ordering"); // remove the original contingency table and replace with transpose myr["Contingency Table"][CrossTabBox(1)] << Delete; myr["Contingency Table"] << Append(mycopy); // show original side by side obj = Contingency( Y( :Age ), X( :sex ) ); Report(obj)["Contingency Table"] << Set Title("Contingency Table - Current output"); myr["Contingency Table"] << Set Title("Contingency Table - Wish List for JMP15"); Is anyone else bothered by the orientation of the contingency table and mosaic plot? I do not use it, but one of our JMP users wanted to flip the orientation of the table so that it matches the mosaic plot. Found out that if you swap X and Y, both the table and mosaic plot update. See script below to run an example and compare them side-by-side. It seems you would want the numbers in the table to line up with the cells in the mosaic plot for easier viewing and referencing. I could not find any previous discussions. It appears the contingency table cannot be changed, it is a CrossTabBox, which does not lend itself to changing anything. The script below has in the top part the work around, which is to make a clone of the table made when switching the x and y, and then appending that to the mosaic plot. Also had to do a little with Value Ordering to get it to line up. Will add to the JMP15 wishlist. I did discuss at JMP Discovery Summit with the EA lab and the developers.     Found some posts on R language, where you can use the dir() argument and specify the horizontal and vertical, while the "count" output in RStudio stays the same for the orientation. # R Mosaic Plot Example - Changing Direction getwd() employee <- read.csv("Products.csv", TRUE, sep = ",",                      na.strings = TRUE) count <- table(employee$EnglishCountryRegionName,                employee$Color) count mosaicplot(count, main = "Countries Mosaic Plot",            sub = "Product Colors by Country",            xlab = "Colors",            ylab = "Countries",            las = 1,            dir = c("h", "v"),            color = "skyblue2",            border = "chocolate") ... View more

Would be nice to have more flexible column properties for spec limits and control limits. Current column properties allow only one spec limit or control limit. But many SPC users will have changing control limits and spec limits over time. Currently Control Chart Builder can use a Phase column to separate calculated control limits. Saving to a table it will show the _LCL, _Mean, _UCL, etc. for each phase, with phase as an additional column. Could spec limits and control limits work in this manner? The ability to use a table to pull in the requirements, or save from Control Chart Builder to the Column and not just the table? See wish list requests for more flexible Value Ordering and Value Labels column properties that may interact with JMP tables. Not sure how it would work with all platforms, such as Process Screening. Currently have to write scripts to add changing control limits or spec limits to charts and other platforms. ... View more

Currently in the Fit Least Squares platform, you can "Show Prediction Expression" under "Estimates". It is shown using coded levels, and harder to interpret for many users for coefficients. Currently our users are told to "Save Columns" >  "Prediction Formula" to a column, then select the formula, and use the menu item "Simplify" to undo coded levels and show as typical coefficients for the model. Any way to add the "Simplify" function to the Prediction Expression when shown in the Fit Least Squares report? (By the way, the "simplify" in the Formula Editor was an awesome addition). Example from the Custom DOE class:   "210 + 186* ( (Thickness - 5)/2.5)"  becomes "-162 + 74.4 * Thickness" using Simplify   The Y intercept of 210 is where the coded level of -1 to +1 is 0, so a positive 210 based on coded levels. But when using Fit Y by X, the equation is -162+74.4*Thickness, which has a negative value for Y intercept when Thickness is zero. ... View more

dt:mycol << Value Labels(mylist) << Use Value Labels( 1 ); For platforms such as Control Chart Builder, JMP allows importing and exporting spec limits, control limits, etc, from a JMP file. Would like to replicate this kind of function for Column Properties such as Value Labels, Value Ordering, etc. Currently have to manually type it in. We have created JSL snippets to export and import from a JMP table frequently used ones like Value Labels. The table has columns named "Values" and "Labels", so if the user is always having to do Value Labels, can use the same. Adding a new Value Label manually, can then save and update that table. Could follow the spec limit approach and use "_Values" and "_Labels" for the column names. Would be nice to do it from the Column Properties menu, but could also be under Column Utilities. Spec limits and Control Limits under Column Properties could also be upgraded if able to do it from the Column Properties user interface. For JMP15 Wish List.   ... View more

 Metrology on semiconductor wafers is necessary to understand process targeting and sources of variation. For new device introduction and initial volume production, wafer metrology may use a large number of sites for establishing targets, setting up APC (Automated Process Control), and determine sources of variation across the wafer. However, ongoing metrology on production is often viewed as non-value added. As manufacturing volume increases and the process is maturing, the question arises: what is the purpose of the continuing metrology in relation to the number of sites? Can the number of sites be reduced to increase throughput at metrology? In the past, sites may have been removed based on current maps with reduced sites on other production, and the impact of the new map analyzed with the current site data. Utilizing the built in JSL function NChooseKMatrix(n,k) which generates all possible combinations for Choose K sites from N sites, and the Prediction Profiler, JMP scripts were created that could quickly analyze and weigh all combinations of varying K, and allow the user to pick the best reduced site map based on various requirements. Examples for different scenarios will be shown.   ... View more

Hi JMP users. When using the JMP Query Builder, we have fast response for SQL servers using "ODBC Driver 11 for SQL Server", but slow response on some Oracle servers using "Oracle in OraClient11g", on the order of 2 to 3 minutes. We discovered by accident that it works just fine using the Custom SQL() argument in the New SQL Query() function in the script, instead of using the generated script from JMP Query Builder.   Has anyone else seen slow response from Oracle ODBC?   Part of the problem as I understand it is that our log in account for this Oracle database does not allow viewing some tables in some schemas on the server, so the query builder takes time when first loading up to by-pass them. You may not be able to replicate if your database access includes all schemas and tables. Submitting direct Custom SQL bypasses this scanning of the schemas and tables?   In the example below, using Query Builder, selected columns and a filter. Took 147 seconds to run for a million rows. Now from Query Builder, select the SQL from the SQL tab, or use the "Convert to custom SQL" menu item and copy the Custom SQL (or run the query from the Custom SQL window), and paste it in as the argument for Custom SQL(), and it took only 7 to 8 seconds for a million rows. // this is the JSL script generated by JMP Query Builder and saved to the datatable // takes ~ 147 seconds to run t1 = Tick Seconds(); New SQL Query( Version( 130 ), Connection(mycon), QueryName( "Test1_147s" ), Select( Column( "X1", "t1" ), Column( "X2", "t1" ), Column( "X3", "t1" ), Column( "X4", "t1" ), Column( "LAST_UPDATE_DATE", "t1", Numeric Format( "m/d/y h:m:s", "0", "NO", "" ) ) ), From( Table( "MY_DATABASE_TABLE", Schema( "SCHEMA" ), Alias( "t1" ) ) ), Where(EQ(Column( "X2", "t1" ),"AAAA",UI( Comparison( Base( "Categorical" ) ) ))) ) << Run foreground; t2 = Tick Seconds(); t2-t1; // 147 seconds // Now just copy the SQL code and put into the CustomSQL argument // takes ~ 8 seconds to run t1 = Tick Seconds(); New SQL Query( Connection(mycon), QueryName( "Test2_8s" ), CustomSQL( "SELECT t1.X1, t1.X2, t1.X3, t1.X4, t1.LAST_UPDATE_DATE FROM SCHEMA.MY_DATABASE_TABLE t1 WHERE ( ( ( t1.X2 = 'AAAA' ) ) )" ) ) << Run foreground; t2 = Tick Seconds(); t2-t1; // 8 seconds   For our JMP Query Builder best practices, will now have to show how to run and save the query and datatable from the Custom SQL window for Oracle databases. ... View more