修正的LARS算法和lasso
写得很好,谢谢分享。跑题一句:这里“皇家序列B”似乎又不如JRSSB直接了,不知道Google能否搜到什么是皇家序列。
[未知用户] 这个可能是习惯性问题,学校老师经常这么叫这么叫的,我们也跟着这么叫了。哈哈,我这就改过来。
8 天 后
Sorry for the difficulty to input chinese.
I have a suggestion for COS. Most articles I saw here are so mathematical, or just a chinese translation of a English paper. I don't think this is quite valuable with regard to the development of COS - why not just read the english paper? I am not saying this article is not good. It is just the question as "why this article is useful to COS, or, to its readers".
I have a suggestion for COS. Most articles I saw here are so mathematical, or just a chinese translation of a English paper. I don't think this is quite valuable with regard to the development of COS - why not just read the english paper? I am not saying this article is not good. It is just the question as "why this article is useful to COS, or, to its readers".
[未知用户] 你说得很好,如果我们的作者都能这样想就好了。我也希望大家能多写一些自己的想法。
[未知用户] 我说的是JRSSB,你正文中漏了个字母J。
不知道作者是否能够分享在R中实现LAR的程序呢?
R的包里有lars包,安装后可以直接调用这个函数。
[未知用户] 这种情况下你可以继续修改你的文章,把程序实现加进去,这样这篇文章就更完整了,有理论有实现,读者读了知其然知其所以然。
1 个月 后
soultion path 的那个向量表达式是怎么得出来的?
[未知用户] 三个方程联立:第一个式子是[latex]X_A'u_A=a1_A[/latex],表示[latex]u_A[/latex]是向量张成空间的角平分线。第二个式子[latex]u_A[/latex]的内积为1,代表单位向量,第三个式子[latex]u_A=X_Ab[/latex],表示[latex]u_A[/latex]是由自变量[latex]X[/latex],线性组合而成。三个方程联立,求解就可以了。
11 天 后
[未知用户] I think such articles still have contributions.
I find the popular statistical topics in China is generally outdated than US. Topics like multivariate analysis, nonparametrics used to be hot 10 years ago, but nowadays the cutting edge focuses on sparse data analysis. Techniques such as variable selection, FDR (false discovery ratio), random matrix have been developing rapidly recently, and it is very helpful if some one can introduce such techniques to the general audience. I am sure they will be popular in China in a few years.
LASSO is a very good topic in such areas, and I am very willing to see if the author can incorporate more interpretation using R.
I find the popular statistical topics in China is generally outdated than US. Topics like multivariate analysis, nonparametrics used to be hot 10 years ago, but nowadays the cutting edge focuses on sparse data analysis. Techniques such as variable selection, FDR (false discovery ratio), random matrix have been developing rapidly recently, and it is very helpful if some one can introduce such techniques to the general audience. I am sure they will be popular in China in a few years.
LASSO is a very good topic in such areas, and I am very willing to see if the author can incorporate more interpretation using R.
[未知用户] 嗯,incorporate more interpretation using r,这一点也是我想说的。
1 个月 后
你好,想问一下你有没有下面的文章
Li, R. (2000), “High-Dimensiona l Modeling via Nonconcave Penalized Like-
lihood and Local Likelihood,” unpublished Ph.D. dissertation, University
of North Carolina at Chapel Hill, Dept. of Statistics.
Li, R. (2000), “High-Dimensiona l Modeling via Nonconcave Penalized Like-
lihood and Local Likelihood,” unpublished Ph.D. dissertation, University
of North Carolina at Chapel Hill, Dept. of Statistics.
7 个月 后
[未知用户] R的包是什么?在哪里可以下载呢?我刚入门,还请多多指教
1 年 后
你好,lasso改进算法的elasticnet包,里面有些参数如actions、allset、beta.pure、vn、mu、normx、meanx、lsigma2等的意思不太明白,希望你多多指教
5 个月 后
文后模拟的代码找不到404
2 个月 后
我可不这么想。如果是学这个统计专业的,当然要看英文、写英文。但是不是统计专业的就完全是另外一回事了,说到底统计方法还的用于实际行业啊。
1 年 后
请问直接调用lars包怎么求线性模型的参数beta
4 年 后
请问Least angle regression文章中的五六节的证明有人会么