Classification of behaviour in housed dairy cows using an accelerometer-based activity monitoring system

Table 1 Confusion matrix obtained for the classification of dairy cow behaviour with the decision-tree classification algorithm

Observed behaviour	Predicted behaviours			Total number of observations
1-min window	Lying	Standing	Feeding
Lying	74.09	22.27	3.64	988
Standing	8.96	82.08	8.96	279
Feeding	0.14	4.21	95.65	736
				2003
5-min window
Lying	74.09	21.76	4.15	193
Standing	5.77	88.46	5.77	52
Feeding	0.0	2.56	97.44	156
				401
10-min window
Lying	77.42	18.28	4.30	93
Standing	4.0	88.0	8.0	25
Feeding	0.0	1.22	98.78	82
				200

The results were obtained using a 1-min, 5-min and 10-min window size and with values of 0.0413 and −0.055 g for threshold A and B, respectively. The values given in the main part of the table correspond to the percentage of observations classified for each behaviour. The final column gives the total number of validated observations (data points) for each behaviour
Values marked in bold indicate the classification sensitivity for each behaviour

ISSN: 2050-3385