Using two methods to measure the accuracy of the prediction results of landslide susceptibility analysis
Yi-Ching Chung
Abstract
There are two methods commonly used in measuring the accuracy of the diagnostic systems or models.
Diagnostic systems are all around us and most of them are used to distinguish between two classes of events.? They are used to reveal diseases in people, malfunctions in nuclear power plants, flaws in manufactured products, threatening activities of foreign enemies, collision courses of aircraft, and entries of burglars. ?For them, the first paper discusses about using the ROC curve (Relative Operating characteristic Curve) to provide a precise and valid measure of diagnostic accuracy.
In second paper, the author proposed a method to provide measures of significance of prediction result when the predictions were generated from spatial databases for landslide hazard mapping. ?The SRC (Success Rate Curve) is used to confirm the stability of the models. ?The theory of the PRC (Prediction Rate Curve) is almost the same with the theory of the SRC but its purpose is using to validate the results of some models over other ones.
Both of these two papers measured the AUC (Area Under the Curve) to validate the significance of prediction results of the models.? The AUC is bigger, the model is better.? I will use these two curves to measure the accuracy of the prediction results of landslide susceptibility analysis, and try to discuss that which one is quite suitable to landslide susceptibility analysis.
Reference
Swets, J. A. (1988) Measuring the accuracy of diagnostic systems, Science, 240, 1285-1293.Chang, J. F., Chung., Fabbri, A. G. (2003) Validation of spatial prediction models for landslide hazard mapping, Natural Hazards. 30, 451–472.