Geographic Information Systems: An Introduction, 3rd Edition

11.6 Attribute Data Accuracy

Considerations of accuracy usually focus on the accuracy of geometrical data, although in practice the accuracy of attribute data is equally important. ISO (2001) has proposed that the following elements be used for defining the accuracy of attribute data:

1. Classification correctness—comparison of the classes assigned to features or their attributes to a universe of discourse (e.g., ground truth reference data set)
2. Nonquantitative attribute correctness
3. Quantitative attribute accuracy

Attribute data often form the basis for classification. Classification correctness is determined by comparing the actual classification with the description of each class. Quality will thus be dependent on a good description of the classes and well-trained staff performing out the classification. In some cases, misclassification may involve simple sorting errors: An object of type A is put in class B. Thus a residential building may be classified as a commercial building if the demarcation between the two classifications is unclear. In other cases, the class structure may be faulty, as when there is no class C for objects containing elements of both A and B. The accuracy of classified data is best expressed with a percentage correctly classified; probable misclassification can be expressed with a matrix; deductive estimates are expressed by standard deviation.

Nonquantitative attributes are the same as nominal attributes and can, for example, include items such as type, ...