Feature Selection Using Maximum Feature Tree Embedded with Mutual Information and Coefficient of Variation for Bird Sound Classification

Table 5

Comparison of Kappa, accuracy, and F₁ scores with feature selection methods in Crane dataset.

Evaluation indicator	Classifier	Feature selection method
		Cor	GR	IG	OR	RF	SU	CS	MICV
		Number of features|the highest value
Kappa	J48	73 | 0.72	69 | 0.74	71 | 0.68	70 | 0.72	68 | 0.71	71 | 0.69	22 | 0.68	25 | 0.75
	NB	73 | 0.75	69 | 0.75	73 | 0.75	73 | 0.75	71 | 0.75	73 | 0.75	53 | 0.79	43 | 0.79
	SVM	73 | 0.84	69 | 0.84	73 | 0.84	73 | 0.84	72 | 0.86	73 | 0.84	73 | 0.84	69 | 0.84
	RFs	66 | 0.89	51 | 0.88	69 | 0.89	73 | 0.89	68 | 0.89	63 | 0.88	36 | 0.90	41 | 0.90
Accuracy	J48	73 | 77.00	69 | 78.00	71 | 73.00	70 | 77.00	68 | 76.00	71 | 73.00	22 | 73.00	18 | 79.00
	NB	73 | 79.00	69 | 79.00	73 | 79.00	73 | 79.00	71 | 79.00	73 | 79.00	53 | 81.00	43 | 83.00
	SVM	72 | 87.00	68 | 87.00	73 | 87.00	73 | 87.00	72 | 89.00	73 | 87.00	73 | 87.00	69 | 87.00
	RFs	66 | 91.00	51 | 90.00	69 | 91.00	73 | 91.00	58 | 91.00	63 | 90.00	36 | 90.00	41 | 91.00
F1 score	J48	73 | 0.77	69 | 0.78	71 | 0.73	70 | 0.77	67 | 0.77	71 | 0.73	25 | 0.73	25 | 0.79
	NB	73 | 0.79	69 | 0.79	73 | 0.79	73 | 0.79	73 | 0.79	73 | 0.79	53 | 0.81	43 | 0.82
	SVM	72 | 0.86	68 | 0.87	73 | 0.86	73 | 0.86	72 | 0.88	73 | 0.86	73 | 0.87	69 | 0.86
	RFs	69 | 0.91	51 | 0.89	69 | 0.91	73 | 0.90	68 | 0.90	66 | 0.90	72 | 0.91	41 | 0.91