研究文章
比较先进的预测性能统计和机器学习技术使用巨大的大数据:证据来自蒙特卡洛实验
|
| 模型 |
ρ= 0.25,= 50 |
ρ= 0.25,= 70 |
|
| n= 100/200/400 |
RMSE |
美 |
RMSE |
美 |
| MCP |
1.123 / 1.055/1.027 |
0.908 / 0.848/0.821 |
1.205 /1.069/1.031 |
0.971 /0.858/0.825 |
| E-SCAD |
1.135 / 1.066/1.034 |
0.917 / 0.856/0.827 |
1.195 /1.086/1.040 |
0.961/ 0.872/0.831 |
| Autometrics |
1.316 / 1.091/1.027 |
1.065 / 0.874/0.822 |
1.316 / 1.091/1.042 |
1.065 / 0.874/0.834 |
| FM_PCA |
3.517 / 3.210/2.829 |
2.839 / 2.576/2.260 |
4.493 / 4.305/3.966 |
3.623 / 3.458/3.173 |
| FM_PLS |
1.528 / 1.200/1.090 |
1.235 / 0.963/0.871 |
1.921 / 1.321/1.126 |
1.551 / 1.059/0.901 |
|
| n= 100/200/400 |
ρ= 0.5,= 50 |
ρ= 0.5,= 70 |
| MCP |
1.145 /1.056/1.027 |
0.925 /0.848/0.821 |
1.318 /1.069/1.032 |
1.062 /0.858/0.825 |
| E-SCAD |
1.112/ 1.058/1.030 |
0.898/ 0.849/0.824 |
1.168/ 1.074/1.035 |
0.940/ 0.862/0.827 |
| Autometrics |
1.156 / 1.062/1.027 |
0.931 / 0.853/0.821 |
1.473 / 1.091/1.041 |
1.191 / 0.874/0.833 |
| FM_PCA |
2.583 / 2.053/1.705 |
2.088 / 1.644/1.365 |
3.933 / 3.334/2.700 |
3.174 / 2.677/2.164 |
| FM_PLS |
1.368 / 1.161/1.080 |
1.105 / 0.932/0.864 |
1.595 / 1.248/1.108 |
1.287 / 1.001/0.886 |
|
| n= 100/200/400 |
ρ= 0.9,= 50 |
ρ= 0.9,= 70 |
| MCP |
1.484 / 1.157/1.042 |
1.198 / 0.930/0.832 |
1.764 / 1.261/1.058 |
1.424 / 1.013/0.846 |
| E-SCAD |
1.201 /1.060/1.019 |
0.968 /0.851/0.814 |
1.291 /1.080/1.021 |
1.040 /0.867/0.817 |
| Autometrics |
4.363 / 1.795/1.031 |
3.528 / 1.443/0.825 |
6.589 / 2.501/1.053 |
5.333 / 2.006/0.843 |
| FM_PCA |
1.169 / 1.099/1.075 |
0.943 / 0.883/0.859 |
1.318 / 1.212/1.165 |
1.065 / 0.974/0.932 |
| FM_PLS |
1.138/ 1.078/1.043 |
0.919/ 0.865/0.834 |
1.184/ 1.095/1.053 |
0.959/ 0.880/0.842 |
|
|
|
请注意。大胆的值表示一个更好的预测。
|
