Unravelling intubation challenges: a machine learning approach incorporating multiple predictive parameters

BMC Anesthesiology

Table 4 Performance of different classification algorithms after data balancing

Metric	Balanced Data- Under sample based on the repeated edited nearest neighbor method				Balanced Data- SMOTE
Metric	Accuracy	Precision	Recall	F1-score	Accuracy	Precision	Recall	F1-score
Euclidean (K = 5)	0.84	0.87	0.75	0.80	0.68	0.17	0.37	0.22
Manhattan (K = 5)	0.80	0.83	0.71	0.76	0.67	0.17	0.39	0.23
Cosine (K = 5)	0.80	0.86	0.67	0.76	0.66	0.17	0.38	0.22
Minkowski (K = 5)	0.84	0.87	0.75	0.80	0.65	0.18	0.45	0.25
Euclidean (K = 3)	0.87	0.88	0.82	0.85	0.73	0.13	0.16	0.13
Manhattan (K = 3)	0.82	0.84	0.75	0.79	0.70	0.12	0.17	0.13
Cosine (K = 3)	0.85	0.88	0.78	0.83	0.68	0.12	0.20	0.14
Minkowski (K = 3)	0.87	0.88	0.82	0.85	0.67	0.12	0.20	0.14
Logistic Regression	0.71	0.70	0.60	0.65	0.66	0.17	0.40	0.23
Svm (RBF)	0.80	0.78	0.78	0.78	0.66	0.15	0.32	0.20
Svm (POLY)	0.82	0.84	0.75	0.79	0.66	0.16	0.35	0.21
Random Forest (INFORMATION-GAIN)	0.76	0.78	0.64	0.70	0.76	0.13	0.14	0.12
Random Forest (GINI-INDEX)	0.74	0.77	0.60	0.68	0.73	0.07	0.16	0.10
Decision Tree	0.77	0.79	0.67	0.73	0.65	0.15	0.33	0.21

ISSN: 1471-2253