Table 1.

Details on confusion matrix and performance measurement metrics of three machine-learning models with different window sizes

Window size	50 k			20 k			10 k			5 k
Performance metrics	SVM	RF	DL	SVM	RF	DL	SVM	RF	DL	SVM	RF	DL
Accuracy	80.43%	81.10%	80.54%	81.88%	73.83%	87.14%	84.12%	72.60%	88.14%	81.99%	75.62%	85.35%
Precision	79.67%	84.92%	82.16%	79.66%	76.92%	88.50%	82.97%	77.26%	84.07%	81.73%	78.47%	80.01%
Recall	79.11%	72.55%	76.05%	80.84%	58.23%	86.81%	78.35%	54.04%	93.04%	78.35%	57.49%	92.81%
F1	79.39%	78.25%	78.99%	80.24%	66.28%	87.65%	80.01%	63.59%	88.33%	80.01%	66.36%	85.93%
AUC	90.07%	90.62%	91.64%	91.00%	83.51%	95.03%	94.29%	82.62%	96.49%	91.24%	84.71%	94.58%
Acc std on cross val.	0.019	0.007	0.026	0.010	0.012	0.031	0.019	0.007	0.022	0.047	0.018	0.099
Prediction Correlation	0.762	0.704	0.725	0.730	0.601	0.802	0.787	0.580	0.825	0.736	0.603	0.777
P-value	1.10E−153	1.88E−121	5.80E−114	1.54E−120	1.59E−93	2.71E−188	9.28E−138	4.89E−89	1.35E−216	1.03E−124	6.81E−98	6.95E−176

DL: our Deep Learning model.