【求助】我做multi-logit模型时，几乎每个系数的SIG值都接近1

WARNINGS
the log-likelihood values are approaching 0.there may be a complete separation in the data. the max lilelihood estimates do not exist

请问这是什么情况？谢谢