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 |
σ= 0.1/0.3,= 50 |
σ= 0.1/0.3,= 70 |
|
| n= 100/200/400 |
RMSE |
美 |
RMSE |
美 |
| MCP |
0.313/0.306/0.303 |
0.253 / 0.246/0.242 |
0.321 / 0.307/0.303 |
0.260 / 0.246/0.242 |
| E-SCAD |
0.319 / 0.309/0.304 |
0.258 / 0.248/0.243 |
0.331 / 0.311/0.305 |
0.267 / 0.249/0.243 |
| Autometrics |
0.318 / 0.308 /0.303 |
0.256 / 0.248 /0.242 |
0.339 / 0.313/0.305 |
0.274 / 0.250/0.244 |
| FM_PCA |
3.373 / 3.055/2.648 |
2.723 / 2.452/2.115 |
4.382 / 4.197/3.847 |
3.534 / 3.374/3.078 |
| FM_PLS |
0.399 / 0.327/0.311 |
0.322 / 0.262/0.249 |
0.625 / 0.347/0.317 |
0.504 / 0.278/0.253 |
|
| n= 100/200/400 |
σ= 0.2/0.6,= 50 |
σ= 0.2/0.6,= 70 |
| MCP |
0.627 / 0.613/0.606 |
0.507 / 0.492/0.484 |
0.643 / 0.614/0.607 |
0.520 / 0.492/0.485 |
| E-SCAD |
0.637 / 0.617/0.609 |
0.515 / 0.496/0.486 |
0.659 / 0.621/0.609 |
0.532 / 0.498/0.487 |
| Autometrics |
0.636 / 0.617 /0.606 |
0.512 / 0.496 /0.484 |
0.667 / 0.625/0.610 |
0.548 / 0.501/0.488 |
| FM_PCA |
3.410 / 3.101/2.704 |
2.753 / 2.489/2.160 |
4.412 / 4.233/3.883 |
3.556 / 3.402/3.106 |
| FM_PLS |
0.798 / 0.654/0.623 |
0.646 / 0.525/0.498 |
1.107 / 0.693/0.634 |
0.892 / 0.556/0.507 |
|
| n= 100/200/400 |
σ= 0.3/0.9,= 50 |
σ= 0.3/0.9,= 70 |
| MCP |
0.941 / 0.920/0.909 |
0.761 / 0.739/0.727 |
0.965 / 0.921/0.910 |
0.780 / 0.739/0.728 |
| E-SCAD |
0.954 / 0.926/0.913 |
0.771 / 0.743/0.730 |
0.985 / 0.930/0.914 |
0.795 / 0.746/0.730 |
| Autometrics |
0.954 / 0.926 /0.909 |
0.768 / 0.744 /0.727 |
1.017 / 0.938/0.916 |
0.823 / 0.752/0.733 |
| FM_PCA |
3.478 / 3.176/2.791 |
2.809 / 2.549/2.230 |
4.467 / 4.281/3.941 |
3.601 / 3.440/3.153 |
| FM_PLS |
1.181 / 0.983/0.935 |
0.956 / 0.789/0.748 |
1.507 / 1.040/0.951 |
1.215 / 0.834/0.760 |
|
|
|
请注意。大胆的值表示一个更好的预测。
|
