Proximity mapping: Weights

With the Weights tab, you can specify proximity (cell) weights for each source and object (case) weights for proximity data. These weights influence how the proximities contribute to the optimization process. This tab is only relevant when the Data type, Proximity Data is selected in the Data tab.

Variables
The Variables list displays all variables in the active dataset. These variables should contain the proximity weights. Select one or more variables and move them to the Weights in Source. These variables must contain the proximity weights for a specific source.
Weights in Source
Weights in Source displays the selected weight variables that are associated with the currently active source. These variables define a matrix of proximity weights that matches the structure of the corresponding proximity matrix. The number of weight variables must be equal to the number of variables in the source matrix.

Use Add Weights, Change Weights, or Remove Weights to manage the weight definitions for the selected source in Sources.

Source(s)
The Source(s) list shows all defined proximity sources to which weights can be assigned. Select a source to view or modify its associated weights.
Object (Case) Weights
With the Object labels field under the selection panes, you can define a variable that contains weights for individual cases (objects). These weights adjust the influence of each object in the loss function when proximity data is used. Case weights are not applicable for multivariate data.
Weights Function

The Weights Function provides three options for special proximity weighting schemes that are functions of the proximity values:

  • None: No special weighting is applied. Only user-specified proximity and object weights (if any) are used.
  • Sammon: Applies weights to the proximity between objects i and j. This weight emphasizes smaller proximities, as in Sammon mapping.
  • Energy: Applies weights that place even more emphasis on smaller proximities.

The total weight applied to each proximity is the product of the user-defined proximity weights, the case weights, and the special weight function. These derived weights govern the relative importance of each proximity in the loss function that is minimized by PROXMAP.