|
Abstract:
|
Spatial classification has received considerable attention in recent decades. In practice, the binary response variable is often subject to measurement error, misclassification. To account for the misclassified response in spatial classification, we proposed validation data-based adjustment methods using interval validation data. Regression calibration and multiple imputation methods are utilized to correct the misclassified responses at the locations where the gold-standard device not available. Generalized linear mixed model and indicator Kriging are applied for spatial prediction at unsampled locations. We perform simulation studies to compare the proposed methods with naïve methods that ignore the misclassification. It is found that the proposed models significantly improve predication accuracy. Additionally, we apply the proposed models for predicting the precipitation area in South Korea.
|
