Table 6 F1 values following imputation of housing dataset, stratified by type and amount missing.
From: Using item response theory as a methodology to impute categorical missing values
Missing type | Num. missing | KNN | MICE\(^1\) ± Std. err | Datawig\(^1\) ± Std. err | IRT |
|---|---|---|---|---|---|
MAR | 535 | 0.328 | 0.203 ± 0.008 | 0.523 ± 0.0004 | 0.523 |
MAR | 1069 | 0.332 | 0.206 ± 0.008 | 0.526 ± 0.0002 | 0.527 |
MAR | 3208 | 0.298 | 0.204 ± 0.004 | 0.515 ± 0.005 | 0.522 |
MAR | 5346 | 0.187 | 0.198 ± 0.002 | 0.438 ± 0.004 | 0.456 |
Marginal means | 0.286 | 0.203 | 0.501 | 0.507 | |
MCAR | 535 | 0.263 | 0.197 ± 0.005 | 0.566 ± 0.003 | 0.563 |
MCAR | 1069 | 0.264 | 0.193 ± 0.005 | 0.545 ± 0.0004 | 0.544 |
MCAR | 3208 | 0.255 | 0.194 ± 0.003 | 0.549 ± 0.0008 | 0.550 |
MCAR | 5346 | 0.258 | 0.194 ± 0.002 | 0.546 ± 0.0006 | 0.553 |
Marginal means | 0.260 | 0.195 | 0.551 | 0.553 | |
