Table 3 Classification accuracy and F1-score of seven classifiers on the first air compressor dataset using different sampling methods. UEDSM consistently achieves the best overall balance, with ALDNet+UEDSM attaining the highest performance.

		ALDNet (%)	SVM (%)	Decision tree (%)	Random forest (%)	Naive Bayes (%)	KNN (%)	XGBoost (%)
UEDSM	Accuracy	98.69	98.44	98.29	98.56	95.58	98.05	98.50
UEDSM	F1-score	84.00	81.87	78.20	82.25	64.05	78.31	80.47
ClusterCentroids	Accuracy	98.23	98.47	64.70	98.53	95.58	98.02	98.07
ClusterCentroids	F1-score	79.92	81.53	15.40	81.81	64.09	77.83	75.62
NearMiss	Accuracy	34.07	17.00	5.66	5.98	6.81	21.67	21.69
NearMiss	F1-score	9.62	7.69	7.19	7.49	7.11	8.59	8.59
TomekLinks	Accuracy	98.63	98.39	98.32	98.63	97.14	98.33	98.54
TomekLinks	F1-score	82.65	81.70	78.41	82.86	72.36	80.18	80.10
SMOTETomek	Accuracy	98.38	98.42	92.48	98.59	96.47	98.12	98.47
SMOTETomek	F1-score	81.59	81.77	43.78	81.96	68.42	78.57	79.49
ADASYN	Accuracy	98.26	98.16	97.23	98.56	95.46	98.03	97.81
ADASYN	F1-score	79.51	79.32	59.52	81.38	63.44	77.82	67.22
KMeansSMOTE	Accuracy	98.48	98.15	98.27	98.61	98.08	98.31	98.55
KMeansSMOTE	F1-score	81.58	78.44	78.03	82.77	77.77	80.06	81.03
BorderlineSMOTE	Accuracy	98.30	98.05	96.45	97.63	95.47	98.01	97.98
BorderlineSMOTE	F1-score	80.63	78.71	34.37	63.62	63.50	77.54	70.49

Quick links

Search