Discriminant Analysis Options
The Discriminant Analysis red triangle menu contains the following options:
Stepwise Variable Selection
(Not available for the Wide Linear method.) Shows or hides the Column Selection control panel. This control panel contains options that enable you to perform stepwise variable selection using covariance analysis and p-values. See Stepwise Variable Selection.
Discriminant Method
Specifies the discriminant method. Choose from Linear, Quadratic, Regularized, and Wide Linear. See Discriminant Method.
Discriminant Scores
Shows or hides a table of the discriminant scores for each row.
Score Options
Provides options for the scoring of the observations.
Show Interesting Rows Only
In the Discriminant Scores report, shows only rows that are misclassified and those with predicted probability between 0.05 and 0.95.
Show Classification Counts
Shows or hides the confusion matrices and confusion rates matrices in the Score Summaries report. A confusion matrix is a two-way classification of actual and predicted responses. A confusion rates matrix is equal to the confusion matrix, with the numbers divided by the row totals. By default, the Score Summaries report shows a confusion matrix and a confusion rates matrix for each level of the categorical X. If you are using JMP Pro with validation, a matrix is given for each set of observations. If you are using JMP with excluded rows, these rows are considered the validation set and a separate Validation matrix is given. See Validation in JMP and JMP Pro.
Show Distances to Each Group
Shows or hides a report that contains each observation’s squared Mahalanobis distance to each group mean.
Show Probabilities to Each Group
Shows or hides a report that contains the probability that an observation belongs to each of the groups defined by the categorical X.
ROC Curve
Shows or hides a Receiver Operating Characteristic (ROC) plot in the Score Summaries report. The plot contains a curve for each level of the response variable. See “ROC Curve” in Predictive and Specialized Modeling.
Precision Recall Curve
Shows or hides a Precision-Recall Curve plot in the Score Summaries report. The plot contains a curve for each level of the response variable. A precision-recall curve plots the precision values against the recall values at a variety of thresholds. See “Precision-Recall Curve” in Predictive and Specialized Modeling.
Select Misclassified Rows
Selects the misclassified rows in the data table and in report windows that display a listing by row.
Select Uncertain Rows
Selects rows with uncertain classifications in the data table and in report windows that display a listing by row. An uncertain row is one whose probability of group membership for any group is neither close to 0 nor close to 1.
When you select this option, a window opens where you can specify the range of predicted probabilities that reflect uncertainty. By default, any row whose probability differs from 0 or 1 by more than 0.1 is defined to be uncertain. Therefore, the default selects rows with probabilities between 0.1 and 0.9.
Save Formulas
Saves distance, probability, and predicted membership formulas to the data table. See Statistical Details for Saved Formulas.
– The distance formulas are SqDist0 and SqDist[<level>], where <level> represents a level of X. The distance formulas produce intermediate values connected with the Mahalanobis distance calculations.
– The probability formulas are Prob[<level>], where <level> represents a level of X. Each probability column gives the posterior probability of an observation’s membership in that level of X. The Response Probability column property is saved to each probability column. For more information about the Response Probability column property, see Using JMP.
– The predicted membership formula is Pred <X> and contains the “most likely level” classification rule.
– The Wide Linear method also saves a Discrim Data Matrix column containing the vector of covariates and a Discrim Prin Comp formula. See Wide Linear Discriminant Method.
Note: For any method other than Wide Linear, when you Save Formulas, a RowEdit Prob script is saved to the data table. This script selects uncertain rows in the data table. The script defines any row whose probability differs from 0 or 1 by more than 0.1 as uncertain. The script also opens a Row Editor window that enables you to examine the uncertain rows. If you fit a new model (other than Wide Linear) and select Save Formulas, any existing RowEdit Prob script is replaced with a script that applies to the new fit.
Make Scoring Script
(Available only in JMP Standard.) Creates a script that constructs the formula columns saved by the Save Formulas option. You can save this script and use it, perhaps with other data tables, to create the formula columns that calculate membership probabilities and predict group membership.
Publish Probability Formulas
(Available only in JMP Pro.) Creates probability formulas and saves them as formula column scripts in the Formula Depot platform. If a Formula Depot report is not open, this option creates a Formula Depot report. See “Formula Depot” in Predictive and Specialized Modeling.
Canonical Plot
Shows or hides the Canonical Plot. See Canonical Plot and Canonical Structure.
Canonical Options
Provides options that affect the Canonical Plot and the Canonical 3D plot.
Show Points
Shows or hides the points in the Canonical Plot and the Canonical 3D Plot.
Show Means CL Ellipses
Shows or hides 95% confidence ellipses for the mean of each group in the Canonical Plot and the Canonical 3D Plot, assuming normality.
Show Normal 50% Contours
Shows or hides a 50% prediction ellipse or ellipsoid for each group. In the Canonical Plot, each ellipse depicts a region in the space of the first two canonical variables that estimates where 50% of the observations for each group should fall, assuming multivariate normality. In the Canonical 3D Plot, each ellipsoid depicts a region in the space of the first three canonical variables that estimates where 50% of the observations should fall, assuming multivariate normality.
Show Biplot Rays
Shows or hides the biplot rays in the Canonical Plot and the Canonical 3D Plot. The labeled rays show the directions of the covariates in the canonical space. They represent the degree of association of each covariate with each canonical variable.
Biplot Ray Position
Enables you to specify the position and radius scaling of the biplot rays in the Canonical Plot and in the Canonical 3D Plot.
– By default, the rays emanate from the point (0,0), which represents the grand mean of the data in terms of the canonical variables. In the Canonical Plot, you can drag the rays or use this option to specify coordinates.
– The default Radius Scaling in the canonical plots is 1.5, unless an adjustment is needed to make the rays visible. Radius Scaling is done relative to the Standardized Scoring Coefficients.
Color Points
Colors the points in the Canonical Plot and the Canonical 3D Plot based on the levels of the X variable. Color markers are added to the rows in the data table. This option is equivalent to selecting Rows > Color or Mark by Column and selecting the X variable. It is also equivalent to right-clicking the graph and selecting Row Legend, and then coloring by the classification column.
Show Canonical Details
Shows or hides the Canonical Details report. See Show Canonical Details.
Show Canonical Structure
Shows or hides the Canonical Structures report. See Show Canonical Structure. Not available for the Wide Linear discriminant method.
Save Canonical Scores
Creates columns in the data table that contain canonical score formulas for each observation. The column for the kth canonical score is named Canon[<k>].
Tip: In a script, sending the scripting command Save to New Data Table to the Discriminant object saves the following to a new data table: group means on the canonical variables; the biplot rays with 1.5 Radius Scaling of the Standardized Scoring Coefficients; and the canonical scores. Not available for the Wide Linear discriminant method.
Canonical 3D Plot
Shows a three-dimensional canonical plot. This option is available only when there are four or more levels of the categorical X. See Example of a Canonical 3D Plot.
Specify Priors
Enables you to specify prior probabilities for each level of the X variable. The following options are available for specifying priors:
Equal Probabilities
Assigns equal prior probabilities to all groups. This is the default.
Proportional to Occurrence
Assigns prior probabilities to the groups that are proportional to their frequency in the observed data.
Other
Enables you to specify custom prior probabilities.
Consider New Levels
Specifies that some points might not fit into any known group and should be considered to be from an unscored new group. See Consider New Levels.
Show Within Covariances
Shows or hides these reports:
– A Covariance Matrices report that gives the pooled-within covariance and correlation matrices.
– For the Quadratic and Regularized methods, a Correlations for Each Group report that shows the within-group correlation matrices.
For each group, the log of the determinant of the within-group covariance matrix (LogDetCov) is also shown. If the determinant is zero, the LogDetCov value is calculated using the part of the covariance matrix that is invertible.
– For the Quadratic discriminant method, adds a Group Covariances outline to the Covariance Matrices report that shows the within-group covariance matrices.
Show Within Covariances is not available for the Wide Linear discriminant method.
Show Group Means
Shows or hides the Group Means report that provides the mean of each covariate. Means for each level of the X variable and overall means appear.
Save Discrim Matrices
Saves a script called Discrim Results to the data table. The script is a list of the following objects for use in JSL:
– a list of the covariates (Ys)
– the categorical variable X
– a list of the levels of X
– a matrix of the means of the covariates by the levels of X
– the pooled-within covariance matrix
Save Discrim Matrices is not available for the Wide Linear discriminant method. See Save Discrim Matrices.
Scatterplot Matrix
(Not available for the Wide Linear discriminant method.) Opens the Scatterplot Matrix platform in a separate window that contains a lower triangular scatterplot matrix for the covariates. The scatterplots include all observations in the data table, even if validation is used. Ellipses with 90% coverage are shown for each level of the categorical variable X. For the Linear discriminant method, these are based on the pooled within covariance matrix.
The options in the Scatterplot Matrix red triangle menu are described in Essential Graphing.
Profiler
Shows or hides an interactive profiler report where the categorical probabilities are combined into a single profiler row. Changes in the factor values are reflected in the estimated classification probabilities. For more information about the options in the red triangle menu, see “Prediction Profiler Options” in Profilers.
See “Local Data Filters in JMP Reports”, “Redo Menus in JMP Reports”, “Group Platform”, and “Save Script Menus in JMP Reports” in Using JMP for more information about the following options:
Local Data Filter
Shows or hides the local data filter that enables you to filter the data used in a specific report.
Redo
Contains options that enable you to repeat or relaunch the analysis. In platforms that support the feature, the Automatic Recalc option immediately reflects the changes that you make to the data table in the corresponding report window.
Platform Preferences
Contains options that enable you to view the current platform preferences or update the platform preferences to match the settings in the current JMP report.
Save Script
Contains options that enable you to save a script that reproduces the report to several destinations.
Save By-Group Script
Contains options that enable you to save a script that reproduces the platform report for all levels of a By variable to several destinations. Available only when a By variable is specified in the launch window.
Note: Additional options for this platform are available through scripting. Open the Scripting Index under the Help menu. In the Scripting Index, you can also find examples for scripting the options that are described in this section.