Visualization in data mining is not a trivial process and if it isn’t performed correctly can produce erroneous results that could mislead and impede further analysis. This section gives a brief overview of visualization concepts that are critical.

Representation

Data objects can be represented based on their attributes’ values if only a single attribute is being considered or they can be represented with the use of rows or columns of a table or as a point in 2D or 3D space.
Attributes on the other hand can be mapped in a similar manner, while categorical attributes require a different form of representation. Nominal attributes are a little bit trickier and require that order not be inferred by the display representation being used. 
Relationships can be mapped implicitly when attributes and data objects are mapped to graphical elements. This is not a trivial process because it is difficult to guarantee that a mapping of attributes and data objects will automatically result in a relationship being mapped in a understandable format. This presents a challenge and there is no optimal solution to this problem but instead one need to choose a visualization technique that will make relationships of interest observable.

Arrangement

Arrangement is another aspect and is crucial because relationships can be deduced from properly arranged visual information. For an example and more information consult the novice track for this section.

Selection

Selection is the process of selecting a subset of objects and attributes for further analysis while discarding or de-selecting others. This is a vital part of visualization because of the curse of dimensionality which makes the visualization of dataset with many attributes cumbersome.