After the SOM has been trained, it is important to know whether it has been properly trained or not, because an optimal map for the given input data should exist. Although several map quality measures have been proposed, the SOM quality is usually measured with two evaluation criteria: resolution and topology preservation. Quantization error is the average distance between each data vector and its BMU for measuring map resolution, and topographic error represents the proportion of all data vectors for which first and second BMUs are not adjacent for the measurement of topology preservation. Thus, this error value is used as an indicator of the accuracy of the mapping in the preserving topology.
Was this article helpful?