AveragingMethodEnum

AveragingMethodEnum in the ABBYY FineReader Engine API — Defines the method (macro or micro averaging) used to compute average classifier performance scores (accuracy, precision, recall, F-measure) for multi-category classifiers.

AveragingMethodEnum enumeration constants are used to define the method used to calculate average performance scores (accuracy, precision, recall, and F-measure) for classifiers with more than 2 categories.

typedef enum {
  AM_Macro,
  AM_Micro
} AveragingMethodEnum;

Elements

Name	Description
AM\_Macro	The scores are calculated for each category separately, and then average over all the categories is taken. Recommended when all categories contain roughly the same number of objects and should influence the resulting score equally.
AM\_Micro	The TP, TN, FP, and FN (true positive, true negative, false positive, and false negative) values for each category are added up, and the scores are calculated from the total values. Recommended when one of the categories is significantly larger than the others, and the resulting score should be more sensitive to the performance on that category.

Used in

IValidationParams::AveragingMethod