Performance Evaluation of Machine Learning Algorithm for Classification of Unintended Pregnancy among Married Women in Bangladesh

Table 3

Performance indicators of all five machine learning algorithms to predict pregnancy intention among married women in Bangladesh.

	LR	RF	SVM	KNN	NB	ENR
Training data set
Accuracy (%)	76.99	78.00	75.98	76.11	77.22	77.12
95% CI	(73.90, 79.88)	(74.95, 80.84)	(72.85, 78.92)	(72.98, 79.04)	(72.37, 81.58)	(74.03, 80.00)
	0.2319	0.1830	0.1629	0.1147	0.1302	0.1600
Sensitivity (%)	23.74	13.64	16.67	10.10	10.52	13.13
Specificity (%)	94.77	99.49	95.78	98.15	99.61	98.48
PPV (%)	60.26	90.00	56.90	64.52	88.89	74.29
NPV (%)	78.82	77.53	77.49	76.58	76.90	77.25
Testing data set
Accuracy (%)	79.29	77.81	76.92	77.22	78.00	77.51
95% CI	(74.57, 83.48)	(73.00, 82.13)	(72.06, 81.31)	(72.37, 81.58)	(74.95, 80.84)	(72.68, 81.86)
	0.3144	0.1623	0.2128	0.1400	0.1753	0.2005
Sensitivity (%)	29.76	11.91	21.43	10.71	12.62	17.86
Specificity (%)	95.67	99.61	95.28	99.21	99.83	97.24
PPV (%)	69.44	90.91	60.00	81.82	96.15	68.18
NPV (%)	80.46	77.37	78.57	77.06	77.39	78.16

PPV = positive predictive value, NPV = negative predictive value.