In today’s edition of Resource Spotlight, JMP Senior Technical Writer Michael Crotty highlights the User Community’s JSL Cookbook. Added to the JMP Community a few years ago, the Cookbook contains code snippets to help you when you’re writing JSL applications. Michael shows you how to search and use the JSL Cookbook: Simply grab the JSL, copy it into a new script and append as needed. And, because it’s part of the JMP Community it’s very easy to add to the Cookbook. Finally, Michael shows you how to quickly and easily “Nominate” future Cookbook articles to help others looking for scripting conclusions.   ... View more

Hi all, Is there a date and location for the 2020 Irish JMP User Forum? Thanks! Michael ... View more

Level: Advanced   Michael Crotty, JMP Senior Statistical Writer, SAS Colleen McKendry, JMP Technical Writer, SAS Marie Gaudard, Statistical Consultant, Statistical Consultant   In situations where anomaly detection is the goal of a predictive model, the underlying data often exhibit an imbalanced class distribution. Namely, the anomalous class is significantly smaller than the non-anomalous class. The modeling goal is usually to identify members of the minority class. However, a straightforward application of predictive modeling techniques can result in a biased and inaccurate model. Many techniques have been proposed to address these issues. We seek to guide JMP Pro users in developing predictive models for imbalanced data. We address JMP Pro approaches to classification into an underrepresented class. We first describe general aspects of the imbalanced class problem: bias, performance measures and approaches to addressing the modeling issues. We then discuss the sampling methods we use in our study; these include weighting, under-sampling, over-sampling and the synthetic minority oversampling technique (SMOTE). For several real data sets that exhibit varying class proportions, we compare the fits obtained using these sampling methods in combination with predictive models available in JMP Pro classification platforms. We perform a similar exploration of sampling techniques and predictive models for a limited range of simulated data sets. For the simulated data sets, we attempt to identify the degree of under-representation for which standard models begin to be affected by class imbalance. We also present conclusions about the relative performance of the sampling methods and predictive models.     ... View more

Level: Intermediate Job Function: Analyst / Scientist / Engineer Michael Crotty, JMP Senior Statistical Writer, SAS Marie Gaudard, Statistical Consultant Colleen McKendry, JMP Technical Writer, SAS   Best Invited Paper Finalist   In situations where anomaly detection is the goal of a predictive model, the underlying data often exhibit an imbalanced class distribution. Namely, the anomalous class is significantly smaller than the non-anomalous class. The modeling goal is usually to identify members of the minority class. However, a straightforward application of predictive modeling techniques can result in a biased and inaccurate model. Many techniques have been proposed to address these issues. We seek to guide JMP Pro users in developing predictive models for imbalanced data. We address JMP Pro approaches to classification into an underrepresented class. We first describe general aspects of the imbalanced class problem: bias, performance measures and approaches to addressing the modeling issues. We then discuss the sampling methods we use in our study; these include weighting, under-sampling, over-sampling and the synthetic minority oversampling technique (SMOTE). For several real data sets that exhibit varying class proportions, we compare the fits obtained using these sampling methods in combination with predictive models available in JMP Pro classification platforms. We perform a similar exploration of sampling techniques and predictive models for a limited range of simulated data sets. For the simulated data sets, we attempt to identify the degree of under-representation for which standard models begin to be affected by class imbalance. We also present conclusions about the relative performance of the sampling methods and predictive models. ... View more

Michael Crotty, JMP Senior Statistical Writer, SAS There are many times when an exact measurement is not possible, but a range of values for the measurement is available. Censored data analysis methods enable you to incorporate the information from both types of measurements. This presentation provides an introduction to censored data situations: when they appear, how to handle them, and what happens when you do not handle them appropriately. This presentation includes examples of censoring in univariate and regression settings by using the Life Distribution and Generalized Regression platforms in JMP and JMP Pro 13.   ... View more

Problem You run an analysis in JMP and want to extract some of the statistics from the platform report. Then you want to place those statistics into a custom report window, so that you can keep those after the data table and platform report are closed. Solution You can get a reference to the report layer of the analysis report and use JSL messages to extract statistics out of  In this example, a Bivariate platform report is generated and you get a reference to the report layer (reportbiv). Using that reference, you can extract variables and strings out of the Bivariate report. Then a new custom report window is created and populated with the statistics and strings that were extracted from the report. Finally, the data table is closed, which closes the Bivariate report, but the custom report window stays open. Names Default To Here(1); sd = Open( "$SAMPLE_DATA/Lipid Data.JMP" ); biv = sd << Bivariate( // biv is the analysis layer. Y( :Triglycerides ), X( :LDL ), Density Ellipse( 0.95, {Line Color( {213, 72, 87} )} ), Fit Line( {Line Color( {57, 177, 67} )} ) ); /* Get a reference to the report layer. */ reportbiv = biv << Report; // Make sure the "Bivariate Normal Ellipse" outline // in the Bivariate report is open. You can then see which // content to extract into the Custom report. reportbiv[Outline Box( "Bivariate Normal Ellipse P=0.950" )] << Close( 0 ); // Extract the correlation coefficient. corrvalue = reportbiv[Outline Box( "Bivariate Normal Ellipse P=0.950" )][Number Col Box( "Correlation" )] << Get( 1 ); // Extract the numeric values from the Summary of Fit report as a matrix. sumfit = reportbiv[Outline Box( "Summary of Fit" )][Number Col Box( 1 )] << Get as Matrix; // Extract the Term column of the Parameter Estimates report as a list of strings. term = reportbiv[Outline Box( "Parameter Estimates" )][String Col Box( "Term" )] << Get(); // Extract the Parameter Estimates values as a matrix. parmests = reportbiv[Outline Box( "Parameter Estimates" )][Table Box( 1 )] << Get as Matrix; dlg = New Window( "Custom Report", Outline Box( "Selected Values", /* The Lineup box defines a two-column layout, each of which contains a Text Box. */ Lineup Box( N Col( 2 ), Text Box( "Factor of Interest: " ), Text Box( term[2] ), ), tb = Table Box( /* Display an empty string in the first column and the text in the second column. */ String Col Box( " ", {"Sample Size: ", "Adjusted RSquare: ", "RSquare: ", "Correlation:"} ), /* Insert a 30 pixel x 30 pixel spacer between the columns. */ Spacer Box( Size( 30, 30 ) ), /* Display an empty string in the first column and the statistics in the second column. */ Number Col Box( " ", sumfit[{5,2,1}] |/ corrvalue ) ), /* Insert a 1 x 30 spacer. */ Spacer Box( Size( 0, 30 ) ), Table Box( /* Display the cloned String Col Box and insert the Parameter Estimates and Standard Error values. */ String Col Box( "Term", term ), Number Col Box( "Estimate", parmests[0,1] ), Number Col Box( "Standard Error", parmests[0,2] ) ) ) ); tb << Set Shade Headings( 0 ); // Turn off shaded table headings. tb << Set Heading Column Borders( 0 ); // Turn off table column borders. Close( sd ); // Close the data table. Discussion JMP platform reports contain two layers: the platform layer and the report layer. If you have a JSL object that is a reference to the platform layer, you can send messages to that object that relate to the analysis. These are generally options that are found in the red triangle menu. If you have a JSL object that is a reference to the report layer, you can send messages to that object that relate to the report itself. These messages can extract values or images, change selections, open and close outline nodes, among other actions. See Also This article is based on an   example in the JMP documentation. ... View more

Problem Instead of a full data set, you have a summarized version that contains the five numbers needed to define a box plot. Solution You can use frequencies to mimic a full data set based on the five-number summary. If you enter the five data points in a data table and then create a second column with frequencies that force your data points to be at the appropriate quantiles, you can use that data table as if it were the raw data. For a box plot with endpoints at the 5% and 95% quantiles, you could use the five-point summary 3, 15, 20, 40, 42 with frequencies of 5, 20, 25, 25, and 20, respectively. The following JSL gives an example: Names Default To Here( 1 ); dt = New Table( "Boxplot", Add Rows( 5 ), New Column( "Summary Point", Numeric, "Continuous", Format( "Best", 18 ), Set Values( [3, 15, 20, 40, 42] ) ), New Column( "Freq", Numeric, "Continuous", Format( "Best", 18 ), Set Values( [5, 20, 25, 25, 20] ) ) ); dt << Graph Builder( Variables( Y( :Summary Point ), Frequency( :Freq ) ), Elements( Box Plot( Y, Legend( 7 ) ) ) ); Discussion The summarized values above are for illustration purposes and can be replaced by other values. See Also Discussion Post ... View more

Problem You have a matrix stored in JSL, and you would like to create a column vector that contains the sums of each row of values in the matrix. Solution Use a combination of the V Sum() function and matrix transposes: M = [1 2 3, 4 5 6, 7 8 9]; V = V Sum( M` )`; Show( V ); The above script results in the following output in the Log window: Show( V ); /*: V = [6, 15, 24];   Discussion   See Also   ... View more

Problem You want to create a new data table column that contains values that are based on the evaluation of a formula. The formula uses the value of another column to determine the value in the new column. Solution This solution uses the Big Class sample data table and creates a formula that uses the value of the Age column to determine a resulting category for a new column called Adolescent Phase. If the value of Age is less than or equal to 12, the formula returns "early"; if the value of Age is greater than 12 but less than or equal to 15, the formula returns "mid"; if the value of Age is greater than 15, the formula returns "late". dt = Open( "$SAMPLE_DATA/Big Class.jmp" ); dt << New Column( "Adolescent Phase", Character, Formula( If( :age <= 12, "early", 12 < :age <= 15, "mid", "late" ) ) ); Discussion A formula column can use the values from other columns in the data table, but it does not have to use them. In fact, you can use random number generation JSL functions in formula columns to create random numbers in a data table. Creating formula columns like this is also sometimes known as feature creation. See Also Documentation ... View more

Problem You want to select one of the comparison circles produced by the Compare Means options in the Oneway platform. This is the JSL equivalent of clicking on one of the comparison circles to select it, and it results in all of the circles in the report getting colored according to the results of the means comparison. Solution You can send the Select Group() message to the Oneway object and specify the group corresponding to the circle you want to select.   dt = Open( "$SAMPLE_DATA\Big Class.jmp" ); ow = dt << Oneway( Y( :height ), X( :age ), All Pairs( 1 ), Comparison Circles( 1 ), SendToReport( Dispatch( {}, "Means Comparisons", OutlineBox, {Close( 1 )} ) ) ); ow << Select Group( 12 ); Discussion If you make a selection and then request further options from the Compare Means submenu, you have to send the Select Group() message again to have the selection appear in all of the comparison circles plots. See Also Discussion post Documentation of Comparison Circles   ... View more