Using BY-Group Processing

To subset data into disjoint subsets defined by one or more BY variables, follow these steps:

  1. On the Graph menu, select a plot type.

    IML Studio displays the dialog box for the plot type.

  2. In the Variables tab, select the plot variables.
  3. Click the BY Variables tab.
  4. Select one or more nominal variables and click Add BY.

    IML Studio uses the values of the BY variables to split the data into disjoint subsets.

  5. Click the BY Options tab and specify options for the plots.
  6. Click OK.

IML Studio splits the data into disjoint subsets, called BY groups. Each BY group is plotted on a separate plot. The BY groups are, by definition, mutually disjoint. Consequently, these plots are not dynamically linked to each other.

The dialog box for creating a plot contains the following tabs:

Details of BY-Group Processing

When you create BY-group plots, the following steps occur:

  1. A new variable, _ObsNum_, is added to the current data table.
  2. The observations corresponding to each BY groups are identified.
  3. The observations in each BY group are copied to a new DataObject. The variables that are copied depend on the Individual plots can reference all variables option on the BY Options tab. Observation properties (such as the color and shape of markers) are also copied to each BY group.
  4. The plots are created.

The BY-group plots are not dynamically linked to the original data. Consequently, selections made to the original data are not reflected in the BY groups. However, you can use an action menu to select observations in the original data that correspond to selected observations in a BY-group plot. Press the F11 key to display the action menu in a BY-group plot. When you select the action menu item, IML Studio looks at the values of the _ObsNum_ variable for the selected observations. IML Studio then selects observations in the original data that contain the same values of _ObsNum_.

Using the action menu to select observations is a cumulative process: if an observation in the original data was selected prior to this action, it remains selected after the action. You can clear selections in the data table the usual way: press the ESC key or click in the upper-left cell of the data table.