Proximity mapping: Restrictions
With the Restrictions tab, you can constrain the solution space by specifying attribute variables. These variables actively guide the configuration by restricting the common object space to be a linear combination of the attributes. This enables supervised mapping, which improves interpretability by anchoring dimensions to meaningful external variables.
- None
- No restrictions are imposed. The configuration is determined solely by the (derived) proximities.
- Attributes
- Enables supervised mapping using one or more external variables as attribute constraints. When this option is selected, you can define which variables act as attributes and specify how each is transformed.
- Variables
-
The Variables list on the left displays all available variables in the dataset. Select one or more variables and click the arrow button to move them to Attributes.
Attributes
The right pane displays all selected attribute variables along with the default transformation function. These transformations control how the attributes are scaled or transformed for use in the restricted configuration.
Click Define Transformation to assign a transformation function to the selected attribute. The following options are available:- Linear keeps the attribute on its standardized scale.
- Spline applies a smooth, monotonic or nonmonotonic transformation, using a piecewise polynomial function.
- Ordinal preserves rank order only.
- Nominal treats categories as unordered and fits them optimally. Gives a single transformation (in contrast to nominal transformation for properties).
You can select multiple rows in the Attributes field to define the same transformation to multiple attributes.
The common space is estimated under the constraint that object coordinates lie within the subspace spanned by the transformed attributes. The Current Number of Dimensions field displays the dimensionality of the solution, which is set in the Criteria tab. Always select a larger number of attributes than the number of dimensions.