Table 3.
The Boston Housing data set: MS-O and MS-N refer to the average number of selected original variables and selected noise variables, respectively. FS-O is the number of original variables that are frequently selected at least 95 times out of 100 repetitions. TS-O refers to the number of original variables selected at least once from 100 repetitions.
| Methods | MSPE | MS-O | MS-N | FS-O | TS-O |
|---|---|---|---|---|---|
| piMoM(MAP) | 24.281 (9.01) | 5.05 (0.22) | 0.01 (0.10) | 5 | 6 |
| piMoM(LS) | 24.265 (9.04) | 5.05 (0.22) | 0.01 (0.10) | 5 | 6 |
| peMoM(MAP) | 24.156 (9.02) | 5.02 (0.14) | 0.00 (0.00) | 5 | 6 |
| peMoM(LS) | 24.165 (9.00) | 5.02 (0.14) | 0.00 (0.00) | 5 | 6 |
| g-prior | 26.314 (9.87) | 3.10 (0.44) | 0.00 (0.00) | 3 | 5 |
| lasso | 30.243 (11.82) | 5.07 (0.87) | 7.77 (11.16) | 4 | 8 |
| scad | 33.993 (10.66) | 5.39 (0.57) | 31.60 (28.28) | 5 | 7 |
| mcp | 26.191 (9.87) | 4.66 (0.74) | 0.54 (1.04) | 3 | 6 |
| Marginal | 26.612 (10.16) | 3.74 (0.88) | 0.41 (0.72) | 3 | 7 |
| Joint | 26.385 (10.25) | 3.77 (0.94) | 0.02 (0.20) | 3 | 6 |