Bitcoin Theft Detection Based on Supervised Machine Learning Algorithms

Table 2

Experimental results corresponding to various settings.

			Acc (%)	Recall (%)	Pre (%)	F1 (%)
A	Unsupervised	LOF	89.5	0.0	0.0	—
		OCSVM	82.8	0.5	0.2	0.3
		MDB	89.6	0.5	0.6	0.6
	Supervised	KNN	95.5	62.2	58.7	60.4
		SVM	94.5	0.0	0.0	—
		RF	95.5	56.3	59.1	57.7
		AdaBoost	95.7	52.9	62.3	57.2
		MLP	94.5	0.0	0.0	—
B	Unsupervised	LOF	89.6	0.5	0.6	0.6
		OCSVM	89.7	0.0	0.0	—
		MDB	89.5	0.0	0.0	—
	Supervised	KNN	98.5	84.8	87.4	86.1
		SVM	94.5	0.0	0.0	—
		RF	99.4	92.4	98.1	95.2
		AdaBoost	99.3	93.6	94.7	94.1
		MLP	94.4	0.0	0.0	—
C	Supervised	KNN	97. 6	92. 4	72. 2	81. 1
		SVM	50. 1	90. 1	9. 0	16. 3
		RF	99. 5	95. 9	95. 9	95. 9
		AdaBoost	98. 7	97. 0	82. 6	89. 3
		MLP	95. 9	93. 0	57. 9	71. 4