Table 5 Scenario 5: Single continuous incomplete variable missing not at random, single continuous complete variable, 100 random runs, 55% missing rate
From: Secure distributed multiple imputation enables missing data inference for private data proprietors
Final analysis | Imputation | Performance | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
Technology | Θ bias | Θ SD | Θ rMSE | \(| \widehat{{\rm{y}}}-{\rm{y}}|\) (μ) | \(| \widehat{{\rm{y}}}-{\rm{y}}|\) (σ) | Discr. | \(| \widehat{{\rm{y}}}-{\rm{y}}|\) (μ) | \(| \widehat{{\rm{y}}}-{\rm{y}}|\) (σ) | Time (s) | Net (MB) | |
500 inds. | Python | 0.002 | 1.11 × 10-16 | 0.002 | 0.055 | 0.016 | 0 | 0.783 | 0.604 | 0.013 | N/A |
PyMICE | 0.004 | 1.11 × 10-16 | 0.004 | 0.055 | 0.017 | N/A | 0.783 | 0.604 | 0.023 | N/A | |
SMC-MICE | 0.003 | 9.70 × 10-3 | 0.010 | 0.052 | 0.017 | 0 | 0.783 | 0.604 | 0.040 | 1.700 | |
MHE-MICE | 0.013 | 1.34 × 10-2 | 0.019 | 0.060 | 0.023 | 0 | 0.783 | 0.602 | 180.3 | 11,045 | |
5000 inds. | Python | 0.011 | 1.11 × 10-16 | 0.011 | 0.074 | 0.016 | 0 | 0.797 | 0.618 | 0.089 | N/A |
PyMICE | 0.030 | 0.0 | 0.030 | 0.074 | 0.016 | N/A | 0.797 | 0.617 | 0.030 | N/A | |
SMC-MICE | 0.022 | 8.32 × 10-2 | 0.086 | 0.101 | 0.045 | 0 | 0.797 | 0.618 | 0.224 | 16.359 | |
MHE-MICE | 0.010 | 6.23 × 10-2 | 0.063 | 0.095 | 0.037 | 0 | 0.796 | 0.617 | 1373 | 115,759 | |
