Table 7.

Mean values of the performance measures for the 15 min prediction during the night (only data considered midnight to 6 am).

	Logistic regression	Random forest	Multilayer perceptron
Specificity	90.4%	98.9%	77.1%
Sensitivity	90.4%	20.5%	61.1%
Precision	29.2%	45.0%	50.0%
F2 score	63.9%	23.0%	58.5%
F1 score	43.3%	35.8%	55.0%
AUC score	90.7%	59.7%	69.1%

The models logistic regression and the random forest are exact re-implementations of (14). The Multilayer Perceptron (MLP) as Deep Neural Network (DNN) model was additionally implemented. The used procedure is illustrated in Figure 3. Best performance measures per ML model are marked in bold.