s******5
发帖数: 513 | 1 the model is
y=mu + beta + g + e
where y is a vector with n individuals
beta is a fixed effect
g=(g1,,,,gm) is random effects
e is the random error
question 1:
we assume var(e)=R*square(sigma_e), where R is an known matrix. How to
incorporate this R matrix in the glm code?
question 2:
g=(g1,,,,gm) is random effects;we assume var(g)=D*square(sigma_g), D is a
known n*m matrix.
Is it to split the matrix D by columns as (d1,....,dm), and take individual
column as a random effect independently? Like:
proc glm data=;
class d1 ... dm;
model y= beta d1 d2 ... dm;
random d1 ... dm;
thank you very much!! | s********1
发帖数: 54 | 2 try "statement random"
【在 s******5 的大作中提到】
: the model is
: y=mu + beta + g + e
: where y is a vector with n individuals
: beta is a fixed effect
: g=(g1,,,,gm) is random effects
: e is the random error
: question 1:
: we assume var(e)=R*square(sigma_e), where R is an known matrix. How to
: incorporate this R matrix in the glm code?
: question 2:
|
|
