Table 9.
Listwise
|
Log
|
CART
|
RF
|
MI
|
|||||||
---|---|---|---|---|---|---|---|---|---|---|---|
30% | 50% | 30% | 50% | 30% | 50% | 30% | 50% | 30% | 50% | ||
|
|||||||||||
Linear | 1.359 | 3.478 | .933 | .832 | 1.038 | .960 | 1.066 | 1.624 | .877 | .576 | |
One split | 3.044 | 3.713 | 1.341 | .943 | 1.021 | 1.042 | 4.923 | 2.183 | .664 | .668 | |
Two splits | .979 | .931 | .847 | .986 | .968 | .996 | 1.424 | 2.738 | .638 | .585 | |
Three splits | 3.344 | 2.837 | .811 | .685 | 1.035 | .945 | 1.160 | 1.237 | .686 | .543 | |
|
|||||||||||
Linear | 1.565 | 5.940 | .884 | .569 | 1.047 | .766 | 1.057 | 1.257 | .883 | .385 | |
One split | 6.089 | 8.160 | 1.685 | .829 | .963 | .885 | 4.340 | 1.909 | .533 | .489 | |
Two splits | 1.044 | 1.500 | .610 | .662 | .873 | .967 | 1.590 | 2.708 | .524 | .450 | |
Three splits | 6.295 | 8.569 | .713 | .629 | 1.035 | .882 | 1.238 | 1.994 | .629 | .522 | |
|
|||||||||||
Linear | .860 | .694 | .901 | 1.084 | 1.063 | 1.112 | 1.046 | 1.696 | .810 | .645 | |
One split | .666 | .887 | 1.357 | 1.101 | 1.044 | 1.039 | 3.392 | 1.607 | .551 | .760 | |
Two splits | .933 | .711 | 1.002 | .999 | 1.186 | 1.070 | 1.682 | 1.963 | .754 | .519 | |
Three splits | .704 | .736 | .805 | .910 | 1.199 | 1.096 | 1.301 | 1.204 | .628 | .697 | |
|
|||||||||||
Linear | .981 | 1.069 | .985 | 1.241 | 1.066 | 1.101 | .961 | 1.410 | .736 | .748 | |
One split | 1.155 | .984 | 3.386 | 1.400 | .966 | .911 | 1.357 | 1.171 | .458 | .575 | |
Two splits | 1.831 | 1.544 | 1.843 | 1.419 | 1.056 | .918 | 1.482 | 1.029 | .571 | .449 | |
Three splits | .978 | .655 | .820 | .905 | 1.005 | .997 | 1.048 | 1.212 | .518 | .469 | |
|
|||||||||||
Linear | .884 | .921 | .933 | 1.221 | 1.045 | 1.207 | 1.078 | 1.718 | .747 | .656 | |
One split | .889 | .892 | 2.484 | 1.223 | 1.031 | .925 | 2.488 | 1.187 | .489 | .564 | |
Two splits | 1.534 | 1.372 | 1.496 | 1.263 | 1.082 | .946 | 1.375 | 1.470 | .638 | .454 | |
Three splits | .860 | .782 | .738 | .929 | 1.046 | 1.070 | 1.015 | 1.236 | .553 | .523 |
Note. MSE ratios computed over MSEPrune. MSE = mean squared error; Log = logistic regression; CART = classification and regression trees; RF = random forests; MI = multiple imputation.