第 3 章 一般线性模型
根据2007ShiJian思想,参考QinYongsong给出的计算框架,做一个高维线性模型的模拟,其中\(X_{ij}\)的取值参考2020ChangJY第13页。相关代码见此。
\[Y = X^\mathsf{T}\beta + \epsilon\]
\(n\)表示样本个数
- \(n = 200,400,800\)
\(p\)表示样本维数
- \(p=cn^{index}\)
- \(index=0,0.16,0.24,0.4\)
- \(c = 3,4,5\)
\(X_i \sim N(0,\Sigma_{p}), \Sigma_{p}=(\sigma_{ij})_{p \times p}\)
\(\sigma_{ij}=I(i=j)+0.5I(i \neq j)\)
\(I(x \in \omega)= \left\{ \begin{aligned} \nonumber 1,& x \in \omega,\\ 0,&x \notin \omega.\\ \end{aligned} \right.\)
3.1 覆盖率
p | EL | MEL | p | EL | MEL | p | EL | MEL | |
---|---|---|---|---|---|---|---|---|---|
0 | 3 | 0.949 | 0.949 | 3 | 0.947 | 0.947 | 3 | 0.941 | 0.941 |
0 | 4 | 0.938 | 0.939 | 4 | 0.944 | 0.944 | 4 | 0.944 | 0.945 |
0 | 5 | 0.926 | 0.928 | 5 | 0.936 | 0.938 | 5 | 0.943 | 0.946 |
0.16 | 7 | 0.916 | 0.924 | 8 | 0.934 | 0.942 | 9 | 0.928 | 0.934 |
0.16 | 9 | 0.894 | 0.910 | 10 | 0.918 | 0.922 | 12 | 0.956 | 0.956 |
0.16 | 12 | 0.870 | 0.882 | 13 | 0.926 | 0.938 | 15 | 0.928 | 0.934 |
0.24 | 11 | 0.862 | 0.872 | 13 | 0.894 | 0.910 | 15 | 0.932 | 0.942 |
0.24 | 14 | 0.822 | 0.846 | 17 | 0.890 | 0.904 | 20 | 0.944 | 0.950 |
0.24 | 18 | 0.786 | 0.800 | 21 | 0.848 | 0.866 | 25 | 0.900 | 0.910 |
0.4 | 25 | 0.620 | 0.654 | 33 | 0.716 | 0.752 | 43 | 0.878 | 0.888 |
0.4 | 33 | 0.428 | 0.488 | 44 | 0.526 | 0.582 | 58 | 0.684 | 0.736 |
0.4 | 42 | 0.188 | 0.218 | 55 | 0.412 | 0.474 | 72 | 0.546 | 0.624 |