logistic求助

是 否 +2 论坛币

k人 参与回答

经管之家送您一份

应届毕业生专属福利!

求职就业群

赵安豆老师微信：zhaoandou666

经管之家联合CDA

送您一个全额奖学金名额~ !

立即领取

感谢您参与论坛问题回答

经管之家送您两个论坛币！

+2 论坛币

模型汇总
        -2 对数似然值        Cox & Snell R 方        Nagelkerke R 方
        147.757a                         .117                                       .450

大家帮着看看，这样的拟合度是不太低了呀，Cox & Snell R 方才0.117，这要在解释的时候可怎么说呀？

扫码加我 拉你入群

请注明：姓名-公司-职位

以便审核进群资格，未注明则拒绝

分享0 收藏0 回帖

关键词：logistic ogistic logisti logist logis 模型

hw检验呢?

If your sample size is large, this is OK.

hplcdadong 发表于 2012-4-1 20:44
If your sample size is large, this is OK.

样本将近1000，不知道是小容量还是大容量？如果ok，解释时应该怎么说呢？谢谢了。

Your sample size is not so small.
Generally speaking, as sample size increases, it's very hard to get "good values" for both Cox & Snell R 方  and Nagelkerke R 方. In your case, Nagelkerke R 方 is really not bad , if  the p value of Hosmer and Lemeshow Test>0.05, you have a good fit model.
Due to the above limitation, a lot of publications don't require Cox & Snell R 方  and Nagelkerke R 方 any more. But they do require "area under the ROC curve" for your model's descrimination power.

类似于线性回归的解释，能够解释方差总体的比重，不过还是应看HL检验

靠不会吧 发表于 2012-4-3 12:28
类似于线性回归的解释，能够解释方差总体的比重，不过还是应看HL检验

嗯，谢谢了～～

hplcdadong 发表于 2012-4-3 02:01
Your sample size is not so small.
Generally speaking, as sample size increases, it's very hard to  ...

Thanks for your kindness~~