用SAS怎么实现 利用赤池AIC信息量最小准则 求最优子集回归模型啊

如题，，，

希望高手解答。。。。。。

给你个程序和说明：
proc reg data=jobprof;
   model y=x1 x2 x3 x4/selection = adjrsq cp aic sbc;
run;

A few notes:
(1) depending on which option is specified first on the right hand side of the equation “selection=”, SAS will use that option as the selection method and will report values for all other options specified after the first option. For example, selection=adjrsq cp aic sbc means selecting best subset model using adjusted R square as criterion, report values in order for cp, aic and sbc.
(2) In SAS 9.1 version that we are currently using, SAS doesn’t use AIC and SBC for selecting best subset model. Therefore, we can not  put “aic” or “sbc” as first option for “selection=”.  If we want to use AIC or SBC as selection method, we can specify slection=adjrsq aic sbc to get AIC and SBC value and then conduct selection by ourselves.

谢谢呢，哈哈

puxingrong 发表于 2010-11-4 21:32
给你个程序和说明：
proc reg data=jobprof;
   model y=x1 x2 x3 x4/selection = adjrsq cp aic sbc;

您好，请问cp是什么？

mosquitog 发表于 2014-4-3 15:21
您好，请问cp是什么？

Cp统计量最小准则是自变量选择的一种判别准则，Cp是Mallow's Cp统计量

好方法，牛人

puxingrong 发表于 2010-11-4 21:32
给你个程序和说明：
proc reg data=jobprof;
   model y=x1 x2 x3 x4/selection = adjrsq cp aic sbc;

超级感谢！！