Skip to main content

Table 3 Performance metrics of eight missing data imputation methods for datasets

From: Comparison of the effects of imputation methods for missing data in predictive modelling of cohort study datasets

 

MAE

RMSE

AUC (95% CI)

ALL

  

0.804(0.796-0.812)

Simple

0.8567

2.9266

0.707*^ (0.695-0.719)

Regression

1.0235

3.5548

0.682*^ (0.667-0.697)

EM

0.6579

2.4939

0.730*^ (0.719-0.741)

MICE

0.8285

2.8699

0.720*^ (0.709-0.731)

KNN

0.2032

0.7438

0.769*(0.759-0.779)

RF

0.3944

1.4866

0.777*(0.769-0.785)

CART

0.7183

2.5534

0.726*^ (0.715-0.737)

Cluster

1.1383

1.1383

0.668*^ (0.663-0.683

  1. * indicates P<0.05 for AUC vs ALL, ^ indicates P<0.05 for AUC vs RF