Skip to main content

Table 1 Average RMSE of 100 simulations using the in vivo myocardial infarction dataset for KNN-TN, KNN-CR and KNN-EU

From: Distribution based nearest neighbor imputation for truncated high dimensional data with applications to pre-clinical and clinical metabolomics studies

MNAR/MAR Sample size Group KNN-TN KNN-CR KNN-EU
6%/3% 25 Cases 0.613 (0.072) 0.619 (0.071) 0.786 (0.075)
25 Controls 0.436 (0.054) 0.441 (0.054) 0.607 (0.047)
50 Cases 0.597 (0.045) 0.602 (0.046) 0.776 (0.048)
50 Controls 0.415 (0.032) 0.420 (0.031) 0.600 (0.028)
10%/5% 25 Cases 0.632 (0.099) 0.637 (0.101) 0.810 (0.087)
25 Controls 0.416 (0.052) 0.419 (0.050) 0.555 (0.044)
50 Cases 0.607 (0.073) 0.610 (0.073) 0.809 (0.069)
50 Controls 0.409 (0.034) 0.412 (0.034) 0.556 (0.029)
20%/10% 25 Cases 0.610 (0.108) 0.612 (0.107) 0..701 (0.091)
25 Controls 0.381 (0.059) 0.389 (0.058) 0.498 (0.048)
50 Cases 0.586 (0.083) 0.586 (0.081) 0.699 (0.071)
50 Controls 0.370 (0.053) 0.381 (0.053) 0.499 (0.041)
  1. Total missing was considered at 9%, 15% and 30%, and within each missing, MNAR was greater than MAR