Machine Learning Readmission Risk Modeling: A Pediatric Case Study

Table 4

Average standard deviation positive predictive value (PPV) [%] of SVM, MLP1, MLP2, and NB for decreasing number of folders in the RCV process. no SMOTE = no oversampling correction of class imbalance is done.

nfolds	SMOTE
	SVM	MLP2	MLP1	NB
10	5.52 0.35	3.92 0.09	5.61 0.47	5.28 0.16
5	5.43 0.27	3.98 0.1	5.25 0.14	5.29 0.31
4	5.39 0.1	3.99 0.01	5.29 0.19	5.29 0.07
3	5.48 0.1	3.94 0.03	5.34 0.07	5.4 0.09
	no SMOTE
	SVM	MLP2	MLP1	NB
10	42.22 29.86	6.23 1.53	NA	9.05 1.11
5	32.47 16.63	5.40 0.59	0.00	9.02 0.95
4	45.24 5.35	6.60 1.96	0.00	9.09 1.13
3	45.24 12.14	6.22 0.82	NA	8.90 0.89