BMC Pregnancy and Childbirth

Table 4 Predictive models of performance of Infant mortality as evaluated on the test data

From: Application of machine learning methods for predicting infant mortality in Rwanda: analysis of Rwanda demographic health survey 2014–15 dataset

			Predictive Models
Evaluation Matrix			Logistic Regression		Random Forest		Decision Tree		Support Vector Machine
Confusion matrix			Predicted		Predicted		Predicted		Predicted
			Dead	Alive	Dead	Alive	Dead	Alive	Dead	Alive
	Observed	Dead	3442	2114	4283	1273	4184	1372	3454	2102
	Observed	Alive	2215	3472	493	5194	536	5151	1425	4262
			%		%		%		%
Accuracy			61.5		84.3		83		68.6
Recall			61.1		91.3		91		75
Precision			62.2		80.3		79		67
F1 score			61.6		85.5		84.7		70.7
AUROC			61.5		84.2		83		68.6

AUROC Area under the Receiver Operating Characteristics

Back to article page

ISSN: 1471-2393

Contact us

Submission enquiries: bmcpregnancyandchildbirth@biomedcentral.com
General enquiries: ORSupport@springernature.com