PieChartTableDataModel

PieChartTableDataModel: Basic Charts

The main responsibility of a PieChartTableDataModel is to determine which data variables are used in the pie chart and what role each variable plays. A PieChartTableDataModel's primary roles are

CategoryVariable: Specifies a classification variable whose values determine the number and arrangement of pie slices in the chart. A unique pie element (slice) is produced for each unique classification value or combination of values when other variable roles are specified. Because a pie chart needs to organize the category variable's data values into groups (typically called bins), the CategoryVariable role uses a ClassificationVariable, which is designed to manage and sort categorical values.
ResponseVariable: Specifies an analysis variable whose values determine the size of each pie slice. The role uses an AnalysisVariable or AnalysisVariableList because these classes associate a statistical computation with a variable instance.

The following code fragment graphs energy production by assigning the Category and Response variable roles to variables named EnergyType and Produced:

 
// Create a data source that must contain a string column
// named EnergyType and a numeric column named Produced
   javax.swing.table.TableModel dataTable = <...>;

// Create a data model and attach the data source to it
   PieChartTableDataModel dataModel =
     new PieChartTableDataModel();
   dataModel.setModel(dataTable);
     
// Assign the Category and Response variable roles
// to appropriate variable(s)
   dataModel.setCategoryVariable(
     new ClassificationVariable("EnergyType"));
   dataModel.setResponseVariable(
     new AnalysisVariable("Produced"));

// Create a PieChart and attach the data model to it
   PieChart pieChart = new PieChart();
   pieChart.setDataModel(dataModel);

Example

Basic Requirements for Creating a Pie Chart: Swing-based code, Servlet-based code

See Also: Valid Variable Roles

PieChartTableDataModel: Data Summarization and Statistic Calculation

When a data model is assigned to a PieChart, the chart summarizes the data before displaying it. During summarization

Analysis values are grouped into bins that are arranged by the classification values. For example, if the classification values represent sales territories, each bin represents one of the territories, and there is a bin for each territory.
The statistic (set by the analysis variable's constructor) is applied to the analysis values in each bin. For example, if the analysis values represent regional sales figures and the statistic is Sum, then the sales figures in each bin are totaled.
The PieChart displays the results for each bin as a pie slice whose size represents the computed statistic value.

The following statistics are the default calculations:

If a response variable is not specified, a FREQUENCY count is calculated for each category value.
If a response variable is specified but no statistic is specified, a SUM is calculated for the response values associated with each category value.

The AnalysisVariable constructor can be used to specify a statistic to calcuate. The following code fragment specifies a mean:

 
AnalysisVariable response=new AnalysisVariable(
  "Var1"                            // column for Response role
  ,GraphConstants.STATISTIC_MEAN    // statistic to calculate
);
PieChartTableDataModel dataModel=new PieChartTableDataModel();
dataModel.setResponseVariable(response);

By default, a PieChart summarizes the data, even if the TableModel's data values have already been summarized. When the values in the data source have already been summarized, you can improve performance by turning off data summarization. Do this by setting the StatisticEnabled boolean to false as follows:

 
PieChartTableDataModel dataModel=new PieChartTableDataModel();
dataModel.setStatisticEnabled(false);

Example

Specifying a Statistic in a Pie Chart: Swing-based code, Servlet-based code

PieChartTableDataModel: Numeric Categorizations

By default, a discrete categorization is performed when a numeric data column is specified for the CategoryVariable role. The format parameter on the ClasificationVariable's constructor can be used to specify the type of categorization (or midpointing) operation to use.

For example, for a numeric data column named Sample, the following code fragment formats the numeric values into categories Initial and Final, and then specifies the numeric variable for the Category role:

 
try { com.sas.text.SASUserDefinedFormat.createFormat(
        "value myFormat 1 - 3 = 'Initial' 4 - 6 = 'Final'");
    } catch (java.text.ParseException e) {
        java.lang.System.out.println("error=" + e.getMessage());
      }
ClassificationVariable myFormattedVar=
  new ClassificationVariable("Sample", "myFormat");

PieChartTableDataModel dataModel=new PieChartTableDataModel();
dataModel.setCategoryVariable(myFormattedVar);

Example

Specifying a Numeric Category Variable: Swing-based code

PieChartTableDataModel: Data Subgroups

To identify data subgroups according to the values of a ClassificationVariable, assign the SubroupVariable role to the appropriate classification variable. Each pie slice in the chart is then divided into segments that represent the values of the Subgroup variable.

For example, if the slices in a pie chart represent energy production, a variable named Year might be assigned the Subgroup role. The Subgroup role would cause each pie slice to be subdivided into segments that represent each year's energy production. The assignment would resemble the following:

 
dataModel.setSubgroupVariable(new ClassificationVariable("Year"));

Different colored pie segments are used to represent each unique value of the Subgroup variable.

Example

Displaying Data Subgroups in a Pie Chart: Swing-based code, Servlet-based code

PieChartTableDataModel: Data-Error Values

A PieChartTableDataModel does not support error values in the data model. If a constructor for an AnalysisVariable that is used in the chart specifies high or low error values, those error values are not used in the chart.

PieChartTableDataModel: Multiple Response Variables

Multiple Response variables can be plotted against the same Category variable by adding the variables to an AnalysisVariableList, and then assigning that list rather than an individual variable to the Response role.

The following code fragment assigns the Category role to variable EnergyType, and the Response role to variables named Produced and Consumed:

 
dataModel.setCategoryVariable(
  new ClassificationVariable("EnergyType"));

AnalysisVariableList multiResponse=new AnalysisVariableList(
  new AnalysisVariable[] {
      new AnalysisVariable("Produced"),
      new AnalysisVariable("Consumed")
  } );
dataModel.setResponseVariable(multiResponse);

The specification generates a pie chart for each unique classification value.

Example

Multiple Response Variables: Swing-based code

PieChartTableDataModel: Multiple Charts in Columns and/or Rows

Separate PieCharts can be generated for each value of a ClassificationVariable by assigning either the ColumnVariable or RowVariable role to that variable.

For example, when charting energy production, a variable named Year might be assigned to the Column role so that a separate chart is generated for the production in each year. The assignment would resemble the following:

 
dataModel.setColumnVariable(new ClassificationVariable("Year"));

If the Column role is assigned, charts are aligned horizontally by the variable's values. If the Row role is assigned, charts are aligned vertically.

The Column and Row variable roles can both be assigned to the same graph. In that case, a separate pie chart is produced for each unique pair of (column, row) values, and the plots are displayed in a grid. For example, when charting product sales, a variable named Year might be assigned to the Column role and a variable named Region might be assigned the Row role. In this case, a separate chart is generated for each region's sales in each year.

Example

Specify Columns: Swing-based code, Servlet-based code

PieChartTableDataModel: Valid Variable Roles

The following table summarizes the PieChartTableDataModel variable roles and the type of data each role supports.

Variable Role Class Type Numeric String Multiple Variables

CategoryVariable*
Assigns the variable that determines the number and arrangement of pie slices ClassificationVariable Yes Yes No

ResponseVariable
Assigns the variable that determines the size of pie slices AnalysisVariable Yes No Yes

SubgroupVariable
Divides pie slices into subgroup segments ClassificationVariable Yes Yes No

ColumnVariable
Aligns multiple charts in columns ClassificationVariable Yes Yes Yes

RowVariable
Aligns multiple charts in rows ClassificationVariable Yes Yes Yes

* required variable

Variable Role	Class Type	Numeric	String	Multiple Variables
CategoryVariable* Assigns the variable that determines the number and arrangement of pie slices	ClassificationVariable	Yes	Yes	No
ResponseVariable Assigns the variable that determines the size of pie slices	AnalysisVariable	Yes	No	Yes
SubgroupVariable Divides pie slices into subgroup segments	ClassificationVariable	Yes	Yes	No
ColumnVariable Aligns multiple charts in columns	ClassificationVariable	Yes	Yes	Yes
RowVariable Aligns multiple charts in rows	ClassificationVariable	Yes	Yes	Yes
* required variable

PieChartTableDataModel: Behavior

To generate a graph, a PieChart needs a data column for the CategoryVariable role. If the role is not assigned in the program code, a PieChart selects the first string column found in the PieChartTableDataModel and uses it as its Category variable. If there is no string column in the data, then it selects the first numeric column.

This default behavior occurs only when the Category role is not assigned. If the role is assigned to a variable that does not exist in the data, the PieChart cannot display a graph.

If the ResponseVariable role is not assigned in the program code, then the sizes of pie slices represent a frequency count of the Category values. If the ResponseVariable role is assigned, then the default statistic for the pie slices is the sum of the response values for each category value.

PieChartTableDataModel: Events

A PieChartTableDataModel fires property change events when

any of its properties are changed
any property in a contained model is modified.

For example, if the name property is changed in the ClassificationVariable that has been assigned the CategoryVariable role, the PieChartTableDataModel fires a ProeprtyChangeEvent stating that the CategoryVariable has changed.

PieChartTableDataModel: Error Handling

When variable roles are assigned, the PieChartTableDataModel does not confirm that the variables exist in the TableModel. Thus, assigning a non-existent data column to a variable role may mean that the PieChart cannot display a graph. For example, if the CategoryVariable role is assigned to a Product column but the TableModel does not contain a Product column, then the PieChart fails to produce a graph.

If the CategoryVariable role is correctly assiged to an existing variable but another role is incorrectly assigned, the PieChart displays a graph without the incorrect role. For example, if the CategoryVariable role is correctly assigned but the ColumnVariable role is not, a graph using the Category role is generated, but the Column role is ignored.