How to incorporate random effect which follows multivariate Gaussian distribution in Linear-Mixed-Effect model?

2 次查看（过去 30 天）

显示更早的评论

Mithun Ghosh 2018-2-8

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/381434-how-to-incorporate-random-effect-which-follows-multivariate-gaussian-distribution-in-linear-mixed-ef

评论： Mithun Ghosh 2018-2-11

在 MATLAB Online 中打开

I have a linear mixed effect model as follows:

y_ij=X_ij*b+Z_ij*B_i .

Here B_i~MGP(0,C). Now how can I incorporate this parameterization of covariance matrix into the model? As the model only allows log-Cholesky,full-Cholesky, diagonal etc. parameterization of the matrix C. Also Z here is B-spline basis matrix.

2 个评论
显示无隐藏无

Bernhard Suhm 2018-2-11

Can you clarify what you are referring to with MGP(0,C)? The doc for LinearMixedModel mentions a couple other covariance matrix types than log/full Cholesky, but I assume above is not one of those.

Mithun Ghosh 2018-2-11

Actually I have n-population so my random effect coefficient is following n-gaussian process which is multivariate gaussian process(MGP).Thus I want to parameterize covatiance matrix according to MGP where C is my multivarite covariance kernel.

请先登录，再进行评论。

请先登录，再回答此问题。