. Author manuscript; available in PMC: 2025 Jan 8.

Published in final edited form as: Curr Biol. 2023 Dec 22;34(1):132–146.e5. doi: 10.1016/j.cub.2023.11.065

Table 1. Classifier performance and evaluation.

The table illustrates the confusion matrices generated for the three evaluation sets using three distinct thresholds for physical contact (4 cm, 5 cm, and 6 cm). It also presents corresponding metrics for Accuracy, Precision, Recall, Specificity, and F1 score for each of these distance thresholds. See also Figure 2, Figure S1, Video S1 and Video S2.

4 cm
		Actual (True) values
		Positive	Negative
Predictive values	Positive	984	9
Predictive values	Negative	108	887

5 cm
		Actual (True) values
		Positive	Negative
Predictive values	Positive	967	9
Predictive values	Negative	8	1053

6 cm
		Actual (True) values
		Positive	Negative
Predictive values	Positive	934	31
Predictive values	Negative	1	1054

	4 cm	5 cm	6 cm
Accuracy	0.9411	0.9917	0.9842
Precision	0.9909	0.9908	0.9679
Recall	0.9011	0.9918	0.9989
Specificity	0.9900	0.9915	0.9714
F1 score	0.9439	0.9913	0.9832