Table 2.
Results from the data analysis
| ridg | src | frc | hyb | hybunc | hierbetas | |
|---|---|---|---|---|---|---|
 |
0.620 | 8.745 | 0.781 | 0.620 | 0.601 | 0.559 |
 |
−0.019 | −0.588 | −0.075 | −0.019 | −0.053 | −0.027 |
 |
0.014 | 0.515 | 0.062 | 0.014 | 0.057 | 0.023 |
| Avg. coverage | 0.91 | 1.00 | 0.98 | 0.91 | 0.99 | 0.94 |
 |
3.372 | 33.785 | 4.023 | 3.372 | 4.674 | 3.310 |
is the empirical MSPE from the validation sample of size 100; 
and 
give the range of the estimate of β for each model, Avg. coverage is the proportion of bootstrap-generated prediction intervals for the validation sample which contained the true outcome, and 
gives the average prediction interval length for the validation sample. hybunc is the hybrid estimator without the non-negativity constraint (Remark6.1).