w**s
发帖数: 26 | 1 请教如何run regression by groups excluding obs i in the group to get the
parameter estimates and then get the predicted value for i based on these
estimates. 比如说一组里有20个obs, 那么for the predicted value of obs 1, use
obs 2-20; for obs 2, use obs 1, 3-20. etc.
基本的model是这样的：
proc reg data=test noprint;
model y=x1 x2 x3 /r;
by group;
output out=reg r=residual;
run;
问题是如何run the above regression excluding obs i in the group.
多谢！！！ | o****o
发帖数: 8077 | 2 you can obtain the value easily by using the influence statistics output by
PROC REG
check the manual.
use
【在 w**s 的大作中提到】
: 请教如何run regression by groups excluding obs i in the group to get the
: parameter estimates and then get the predicted value for i based on these
: estimates. 比如说一组里有20个obs, 那么for the predicted value of obs 1, use
: obs 2-20; for obs 2, use obs 1, 3-20. etc.
: 基本的model是这样的：
: proc reg data=test noprint;
: model y=x1 x2 x3 /r;
: by group;
: output out=reg r=residual;
: run;
| w**s
发帖数: 26 | 3 It works!!!太酷了!非常感谢！
by
【在 o****o 的大作中提到】
: you can obtain the value easily by using the influence statistics output by
: PROC REG
: check the manual.
:
: use
|
|
