Table 4.

Imputation performance comparison based on the nRMSE metric for each analyte and imputation method

Analyte	Training set						Test set
	Interp.	KNN	Selected	KNN+Interp.	Interp.+KNN	3D-MICE	Interp.+KNN	3D-MICE
		K = 3	method	K = 3	K = 3		K = 3
Chloride	0.2017	0.1966	KNN	0.1966 (1.6%)	0.1915 (4.1%)	0.1997	0.1921 (4.0%)	0.2000
Potassium	0.2590	0.2539	KNN	0.2539 (2.9%)	0.2505 (4.2%)	0.2614	0.2542 (3.4%)	0.2632
Bicarb.	0.2165	0.2335	Interp.	0.2165 (6.9%)	0.2165 (6.9%)	0.2326	0.2118 (8.5%)	0.2314
Sodium	0.2242	0.2143	KNN	0.2143 (− 0.3%)	0.2113 (1.1%)	0.2136	0.2085 (2.8%)	0.2145
Hematocrit	0.2248	0.1445	KNN	0.1445 (0.1%)	0.1434 (0.9%)	0.1447	0.1518 (-0.9%)	0.1505
Hemoglobin	0.2282	0.1453	KNN	0.1453 (− 1.7%)	0.1451 (− 1.5%)	0.1429	0.1485 (0.2%)	0.1488
MCV	0.2582	0.2626	Interp.	0.2582 (3.6%)	0.2582 (3.6%)	0.2679	0.2644 (2.5%)	0.2713
Platelets	0.1778	0.2304	Interp.	0.1778 (21.3%)	0.1778 (21.3%)	0.2260	0.1794 (21.8%)	0.2294
WBC counts	0.2183	0.2430	Interp.	0.2183 (14.6%)	0.2183 (14.6%)	0.2555	0.2198 (14.1%)	0.2560
RDW	0.2100	0.2402	Interp.	0.2100 (15.8%)	0.2100 (15.8%)	0.2493	0.2056 (16.4%)	0.2458
BUN	0.1521	0.2231	Interp.	0.1521 (17.7%)	0.1521 (17.7%)	0.1848	0.1546 (16.3%)	0.1846
Creatinine	0.2130	0.2307	Interp.	0.2130 (7.0%)	0.2130 (7.0%)	0.2291	0.2135 (8.7%)	0.2338
Glucose	0.2817	0.2717	KNN	0.2717 (1.9%)	0.2683 (3.1%)	0.2769	0.2677 (3.3%)	0.2769
Average	0.2204	0.2223	–	0.2055 (7.4%)	0.2043 (7.9%)	0.2219	0.2055 (8.1%)	0.2235

Best performances are highlighted in bold. The percentage of improvement over 3D-MICE is given in parentheses