最近在做MMNL模型选择的课题,我从Berkeley的Kenneth Train教授的网站上下载了他的用Gauss语言写的Mixed Logit Estimation Routine for Cross-Sectional Data程序,有些地方不太明白,想请教大家。
1、软件包中的data数据显示有1000行,28列,似乎都是分类变量,请问Train教授的这个程序是只针对分类变量写的吗?他可以做含有连续自变量的模型吗?数据中哪几列是因变量,哪些是自变量呢?
2、程序开头有如下一段程序,
@ Number of observations@
NOBS = 3000;
@ Maximum number of alternatives represented in any choice situation@
NALT = 4;
@ Create or load XMAT @
@ XMAT here contains all explanatory variables and the dependent variable @
load XMAT[NOBS,28] = c:\temp2\data.asc;
@ Number of variables in XMAT @
NVAR = 7;
@ Number of the variable in XMAT that is the dependent variable @
IDDEP =7;
请问NALT、NVAR、IDDEP各指什么呢?为什么矩阵XMAT的列数要等于NVAR*NALT 呢?
请高手赐教,不胜感激:)
[此贴子已经被作者于2009-3-19 20:34:19编辑过]