Exploring best-matched embedding model and classifier for charging-pile fault diagnosis

Cybersecurity

Table 15 Precision, recall, and F1-score result (%)

Classifiers	Metric*	GloVe	Word2vec	BERT
RF	P	80.34	82.21	81.41
	R	79.67	81.26	80.32
	F	80.00	81.73	80.86
SVM	P	42.00	44.75	45.47
	R	27.58	31.23	30.47
	F	33.29	36.79	36.49
KNN	P	41.05	42.34	41.19
	R	42.02	43.25	42.72
	F	41.52	42.79	41.94
MLP	P	47.28	33.14	41.96
	R	45.43	33.88	41.31
	F	46.34	33.51	41.63
RNN	P	82.98	78.44	82.06
	R	82.91	78.26	81.85
	F	82.94	78.35	81.96
AB	P	32.94	31.66	28.89
	R	23.75	24.04	22.27
	F	27.60	27.33	25.15
GBDT	P	43.32	44.83	43.91
	R	42.31	41.37	43.84
	F	42.81	43.03	43.88
DT	P	78.96	78.01	78.77
	R	78.90	77.90	78.43
	F	78.93	77.96	78.60
ET	P	76.94	76.86	77.40
	R	76.72	76.72	77.14
	F	76.83	76.79	77.27
VOTE	P	61.23	58.60	60.94
	R	52.80	51.74	56.28
	F	56.70	54.95	58.51

*There are three metrics including precision, recall and F1-score. We use P to denote precision, R to denote recall, and F to denote F1-score