L*****2
发帖数: 66 | 1 面世时被问到这个问题, 除了link funciton 不同之外,还那些不同,
问到什么时候用logit什么时候用probit,特别是如果response是非常rare的事件,应该
用probit还是logit?
谢谢 | h***i
发帖数: 3844 | 2 用LRT，不过看起来好像logit model尾巴肥一些，发生outlier的概率大一些.
【在 L*****2 的大作中提到】
: 面世时被问到这个问题, 除了link funciton 不同之外,还那些不同,
: 问到什么时候用logit什么时候用probit,特别是如果response是非常rare的事件,应该
: 用probit还是logit?
: 谢谢
| L*****2
发帖数: 66 | 3 hezhi 是说liklihood ratio test 去比叫那个model 好?怎么个比较? | f*******r
发帖数: 257 | 4 In practice, they are equally good. In theory, if there are a lot of
observations in the tail, it may be better to use logit... The coefficient
estimated by probit is about .6 of that by logit.
Rare event case should be estimated by something else, neither logit nor
probit is appropriate:
http://www.stanford.edu/~tomz/software/software.shtml | L*****2
发帖数: 66 | 5 freerider,
Thank you very much for your help!
Can you explain more why it may be better to use logit if there are a lot of
observations in the tail. I know logistic curve is slightly flatter than
probit curve. Is that related to your answer?
Thanks again | f*******r
发帖数: 257 | 6 There are two interpretations of a binary model, one is through link
function, the other one is a utility function that economists like to think
of. Think of purchasing a TV set: all you observe is a customer bought or
not. But the process behind that is a latent one: a utility function that
when reaching some value, the consumer buys. What I meant by observation at
the tail is for this utility function. If you have a large chunk of
utilities at high value or low value (such as consumers who |
|
