Table 9:

Gaussian data with n = 200 and p = 1001 under MNAR. Approximately 40% features and 90% cases contain missing values. Here Linear RR and MICE are not included due to running out of RAM.

Models	Style	Time(s)	Imp MSE	Bias(${\widehat{\beta }}_1$)	CR(${\widehat{\beta }}_1$)	SE(${\widehat{\beta }}_1$)	SD(${\widehat{\beta }}_1$)
SoftImpute	SI	25.1	0.0443	−0.2550	0.52	0.1570	0.2164
GAIN	SI	111.1	0.7395	0.6488	0.18	0.1719	0.5830
MIWAE	SI	53.2	0.1116	0.5249	0.35	0.1902	0.2523
not-MIWAE	SI	45.4	3.981	0.9897	0.0	0.1080	0.1451
Sinkhorn	MI	116.9	0.0889	0.5445	0.38	0.2406	0.2237
MI-NNGP1	MI	4.9	0.0119	0.0351	0.89	0.1194	0.1422
MI-NNGP1-BS	MI	4.9	0.0166	0.0383	0.94	0.1424	0.1416
MI-NNGP2	MI	10.5	0.0085	0.0356	0.91	0.1160	0.1310
MI-NNGP2-BS	MI	9.9	0.0092	0.0343	0.93	0.1263	0.1301
Complete data	-	-	-	0.0350	0.94	0.1122	0.1173
Complete case	-	-	-	0.2824	0.76	0.3447	0.4201
ColMean Imp	SI	-	0.1130	0.7022	0.11	0.2572	0.1941