Fig. 1

Example calucaltion of global accuracy, state-wise sensitivity, specificity, and precision using confusion matrices

Example calculation of state-wise sensitivity, specificity and precision for behaviour 2 (in our case foraging) alongside global accuracy using confusion matrices. The confusion matrix colour fills indicate the idealized distribution of the data, with dark squares in the diagonal representing high concentrations of data correctly predicted and clear squares at the edge indicating low concentrations of incorrectly predicted data. TP (true positive – light green outline): number of predictions where the classifier correctly predicts the positive class as positive, TN (true negative – yellow outline): number of predictions where the classifier correctly predicts the negative class as negative, FP (false positive – red outline): number of predictions where the classifier incorrectly predicts the negative class as positive, FN (false negative – dark green outline): number of predictions where the classifier incorrectly predicts the positive class as negative