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 |
1.167 / 1.078/1.056 |
0.943 / 0.866/0.845 |
1.254 / 1.110/1.065 |
1.012 / 0.892/0.851 |
| E-SCAD |
1.175 / 1.091/1.062 |
0.952 / 0.877/0.850 |
1.241/ 1.124/1.074 |
1.002/ 0.904/0.859 |
| Autometrics |
1.192 / 1.100/1.064 |
0.963 / 0.884/0.851 |
1.392 / 1.126/1.071 |
1.121 / 0.908/0.858 |
| FM_PCA |
3.520 / 3.222/2.858 |
2.848 / 2.589/2.288 |
4.569 / 4.274/3.952 |
3.695 / 3.429/3.165 |
| FM_PLS |
1.568 / 1.231/1.119 |
1.268 / 0.990/0.896 |
1.972 / 1.367/1.166 |
1.591 / 1.101/0.932 |
|
| n= 100/200/400 |
ρ= 0.50,= 50 |
ρ= 0.50,= 70 |
| MCP |
1.324 /1.222/1.185 |
1.073 /0.987/0.949 |
1.448 /1.234/1.197 |
1.177 /0.993/0.957 |
| E-SCAD |
1.318/ 1.238/1.191 |
1.068/ 0.996/0.954 |
1.382/ 1.248/1.206 |
1.122/ 1.005/0.965 |
| Autometrics |
1.330 / 1.222/1.187 |
1.080 / 0.985/0.951 |
1.630 / 1.255/1.202 |
1.318 / 1.011/0.964 |
| FM_PCA |
3.570 / 3.279/2.916 |
2.889 / 2.624/2.333 |
4.607 / 4.247/4.021 |
3.716 / 3.381/3.219 |
| FM_PLS |
1.720 / 1.392/1.258 |
1.389 / 1.121/1.005 |
2.108 / 1.503/1.303 |
1.702 / 1.206/1.042 |
|
| n= 100/200/400 |
ρ= 0.90,= 50 |
ρ= 0.90,= 70 |
| MCP |
2.953 / 2.408 /2.364 |
2.449 / 1.997 /1.936 |
3.608 / 2.538 /2.368 |
2.961 / 2.100 /1.940 |
| E-SCAD |
2.714/2.380/ 2.366 |
2.267/1.976/ 1.937 |
3.039/2.498/ 2.370 |
2.525/2.069/ 1.941 |
| Autometrics |
3.250 / 2.480/2.358 |
2.693 / 2.049/1.930 |
4.273 / 2.594/2.394 |
3.494 / 2.146/1.957 |
| FM_PCA |
4.165 / 3.871/3.563 |
3.387 / 3.126/2.868 |
5.051 / 4.735/4.506 |
4.111 / 3.810/3.609 |
| FM_PLS |
2.941 / 2.579/2.476 |
2.439 / 2.122/2.020 |
3.341 / 2.796/2.544 |
2.749 / 2.293/2.072 |
|
|
|
请注意。大胆的值表示一个更好的预测。
|
