ABSTRACT
This chapter covers the basics of classification accuracy assessment; for more details please read some of the articles in the Additional Readings section. User's accuracy reflects commission errors; of the pixels committed (predicted) to a class, some predictions will be incorrect. A crucial assumption in producing the error matrix, user's and producer's accuracy values, and the KHAT value, is that the reference data are truly representative of the entire classification. Sometimes aerial photography is interpreted as a substitute for field collection of reference data. It is possible to have a poor classification and yet report a high classification accuracy estimate. It is tempting to minimize cost and time by using the data collected from the training field areas as reference data. The error matrix, producer's and user's accuracy and KHAT value have become standard in assessment of classification accuracy.
