c**********e
发帖数: 2007 | 1 When the number of repeated measure is not many, random effect model is good
. I know people use mixed model often in pharmaceutical industry.
But in financial industry, the repeated measure is so many, say unemployment
rate each country in the last 20 years (240 monthly obs). Do we still use
mixed model or random effect model to do panel analysis? Or should we just
use linear regression?
Anybody has experience on this? |
t********6
发帖数: 43 | 2 Mixed model is only problematic when there is too little repeated
measurement, or too little subjects/observations. So I won't worry about.
If you don't have many countries, AR(1) and un-structured
corr matrix will probably have too many parameters to estimate (20yrs
parameter). You can try compound symmetry then. |
A*******s
发帖数: 3942 | 3 什么时候应该用random effect，什么时候random effect estimate趋近fixed effect
estimate，如何从bias variance trade-off的角度来理解random vs. fixed...这都是
我常问的面试问题。lz搞清楚这些问题就有答案了... |
c*****1
发帖数: 115 | 4 我正在用GLMM，还没有想过你的这些问题，能不能简单展开讲讲，万一碰到你的面试也
有了准备不是：）
effect
【在 A*******s 的大作中提到】
: 什么时候应该用random effect，什么时候random effect estimate趋近fixed effect
: estimate，如何从bias variance trade-off的角度来理解random vs. fixed...这都是
: 我常问的面试问题。lz搞清楚这些问题就有答案了...
|
e***7
发帖数: 862 | 5 顶下，同好奇精算兄怎么答
【在 c*****1 的大作中提到】
: 我正在用GLMM，还没有想过你的这些问题，能不能简单展开讲讲，万一碰到你的面试也
: 有了准备不是：）
:
: effect
|
W**********E
发帖数: 242 | 6 没做过这么时间点的longitudinal data，如果有这么多的时间点，你可能要考虑比较
深的方法，什么functional data analysis, smoothing之类的。不懂瞎说。
另外你用linear regression,那你如何解释clustering within each country?
good
unemployment
【在 c**********e 的大作中提到】
: When the number of repeated measure is not many, random effect model is good
: . I know people use mixed model often in pharmaceutical industry.
: But in financial industry, the repeated measure is so many, say unemployment
: rate each country in the last 20 years (240 monthly obs). Do we still use
: mixed model or random effect model to do panel analysis? Or should we just
: use linear regression?
: Anybody has experience on this?
|
