s***r 发帖数: 1121 | 1 Suppose in a regression, I have 1900 independent variables, among which
1890 are dummy variables. So, in the regression output, I do not want it
to show all 1900 parameters as it will take too much space. All I want is
the output of the remaining 10 parameters; not dummy variables.
Is there a SAS statement that can suppress the output of these 1890
parameters (and their t-statistics, standard errors, etc.)
Thank you. 5 bao zi will be given. | D******n 发帖数: 2836 | 2 noprint + outest ?
【在 s***r 的大作中提到】![](/moin_static193/solenoid/img/up.png) : Suppose in a regression, I have 1900 independent variables, among which : 1890 are dummy variables. So, in the regression output, I do not want it : to show all 1900 parameters as it will take too much space. All I want is : the output of the remaining 10 parameters; not dummy variables. : Is there a SAS statement that can suppress the output of these 1890 : parameters (and their t-statistics, standard errors, etc.) : Thank you. 5 bao zi will be given.
| s***r 发帖数: 1121 | 3 I used Noprint already. What I want is not to output regression results for
these dummy variable, only show the regression results for other non-dummy
variables. | k****i 发帖数: 347 | 4 用ods output。输出parameter estimate所在的table,然后在proc print里用where只
选相应的variable name即可 | s***r 发帖数: 1121 | 5 I also used ods output. It is ok, but it is still not what I really want. | k****i 发帖数: 347 | 6 我觉得sas的output是以table为单位的,所以你的要求基本上是不可能滴,呵呵
my 2 cents | z**********i 发帖数: 12276 | 7 用KEEP=VAR1 VAR2...,加在你输出的DATASET后面,怎么样?
for
【在 s***r 的大作中提到】![](/moin_static193/solenoid/img/up.png) : I used Noprint already. What I want is not to output regression results for : these dummy variable, only show the regression results for other non-dummy : variables.
| o****o 发帖数: 8077 | 8 one way I can think of is using Partition Regression where
you first regression your dependent variable and each of the 10 other
independent variables against all dummy variables, obtain their residuals,
then regression the residuals of dependent variable against the residuals of
the 10 independent variables. Theoretically you shall have the same result.
But this only works for OLS, don't know if it applies to more advanced
regression analysis, such as FGLS, etc
【在 s***r 的大作中提到】![](/moin_static193/solenoid/img/up.png) : Suppose in a regression, I have 1900 independent variables, among which : 1890 are dummy variables. So, in the regression output, I do not want it : to show all 1900 parameters as it will take too much space. All I want is : the output of the remaining 10 parameters; not dummy variables. : Is there a SAS statement that can suppress the output of these 1890 : parameters (and their t-statistics, standard errors, etc.) : Thank you. 5 bao zi will be given.
|
|