a*********n 发帖数: 1331 | 1 要用Spatial Multilevel分析:
三层:
第一层: individual events (negative binomial, not Gaussian)
第二层: urban facilities (e.g. grocery stores, bus stops),
第三层: neighborhood conditions, such as unemployment rate and poverty rate.
目的:主要是看哪些因素影响第一层。
注:
1、第一层的individual events是negative binomial分布的,大部分是0。
2、第一层和第二层没有包含关系,都是单个spatial points. 但是第一层的分布由第
二层的分布决定。
建这个multilevel spatial model,该用哪个统计方法?
是否要上Bayesian hierarchical spatial model? 还是HLM就行?
有现成的package或code吗?
多谢。 |
s******a 发帖数: 36 | 2 If the second level is not hierarchically superior to the first level, maybe
they should be combined and only two levels are used. If using STATA,
xtmixed/gllamm may be good options |
C******n 发帖数: 284 | 3 Take a look at Goldstein and Browne's work on cross-classified and multiple
membership model, for example:
Browne, William and Harvey Goldstein. 2010. "MCMC Sampling for a Multilevel
Model With Nonindependent Residuals Within and Between Cluster Units."
Journal of Educational and Behavioral Statistics 35:453-473.
rate.
【在 a*********n 的大作中提到】 : 要用Spatial Multilevel分析: : 三层: : 第一层: individual events (negative binomial, not Gaussian) : 第二层: urban facilities (e.g. grocery stores, bus stops), : 第三层: neighborhood conditions, such as unemployment rate and poverty rate. : 目的:主要是看哪些因素影响第一层。 : 注: : 1、第一层的individual events是negative binomial分布的,大部分是0。 : 2、第一层和第二层没有包含关系,都是单个spatial points. 但是第一层的分布由第 : 二层的分布决定。
|
a*********n 发帖数: 1331 | |
a*********n 发帖数: 1331 | 5 The dependent variable follows negative binomial distribution. xtmixed seems
apply to multilevel linear modeling. xtmixed still working in my project?
maybe
【在 s******a 的大作中提到】 : If the second level is not hierarchically superior to the first level, maybe : they should be combined and only two levels are used. If using STATA, : xtmixed/gllamm may be good options
|
p*********u 发帖数: 219 | 6 没有弄清楚楼主具体的问题,我是搞GIS的有时候也用到模型,如果都可以用图层表达
的话,是否可以考虑用R(里面有很过现成的模型(多种类型的Regression)?有一个
email list:R-sig-Geo 是个很好发问的地方。 |
D***e 发帖数: 21 | |