Assess Accuracy

Description

Generate an accuracy report for a supervised classification application.

Usage

Given the output of a supervised classification model and labeled ground truth data (i.e., testing data), a report is generated summarizing the performance of the model for the locations of the testing data (ideally, locations not used for training the model). Accuracy metrics are chosen to minimize accuracy bias due to imbalanced ground truth classes.

Parameters

Parameter Name	Type	Direction	Data Type	Dialog Reference
Input Testing Raster (*.tif)	Required	Input	Raster Layer	Input testing raster of integer type that represents cells labeled with their corresponding ground truth class value for locations where the supervised classification model was not trained. Must be in TIF format.
Input Prediction Raster (*.tif)	Required	Input	Raster Layer	Raster containing the predictions that are to be assessed. The prediction raster must have the same extents as the testing raster. Must be in TIF format.
Output Metrics Directory	Required	Output	Folder	Directory to store output accuracy assessment items. If the directory does not exist, it will be created.
Optional Accuracy Outputs	Optional	Input	Value Table	Optional inputs to assess the accuracy of probability prediction outputs from the supervised classifier. This option will result in an additional accuracy output stored in the Metrics Directory. Enter the class probability raster and the corresponding class value. Probability raster(s) must be in TIF format.