Table 2 Confusion matrix showing performance of the 5-nearest neighbor classification algorithm trained on data from Fox1 and tested on data from Fox2

From: Use of bio-loggers to characterize red fox behavior with implications for studies of magnetic alignment responses in free-roaming animals

Accuracy (%):	66.7	True class			Precision (%)
Cohen’s kappa:	0.50	Leaping	Foraging	Trotting	Precision (%)
Predicted class	Leaping	7	0	0	100.0
	Foraging	3	5	6	35.7
	Trotting	0	2	10	83.3
	Recall (%)	70.0	71.4	62.5

The results displayed in this matrix can be viewed as estimates of how well the classifier, trained on Fox1, would perform on new, unseen data from a different fox
Italicized values indicate cases where ground-truth behaviors (i.e. 'true class') match classifier prediction (i.e. 'predicted class')

Back to article page