Using Decision Risk and Decision Accuracy Metrics for Decision Making for Remote Sensing and GIS Applications

A confusion matrix or error matrix or contingency table is the most widely used evaluation technique for assessing the difference in actuals with the results of spatial data processing. The confusion matrix provides the basis for computing various metrics such as accuracy, precision, prevalence, sensitivity, and specificity for quantifying the effectiveness of a data classification or a model or an algorithm applied to various applications of remote sensing and Geographic Information System. Since most of these metrics are based on both chance and occurrence but not on the absolute occurrence, in this paper, a set of new metrics is used that are reliably based on the absolute outcomes of the decisions. Decision risk metrics are defined based on errors variable over differentially among different classification groups for decision making. Decision accuracy metrics are defined in terms of decision risk, and absolute accuracy metrics are defined in terms of absolute occurrence that provides with the information for the evaluation of fuzziness in different classification groups and for the comparison among models, algorithms, and datasets.