R 是否可以在多级数据集中使用位置尺度和形状的广义加性模型 (GAMLSS) 并绘制百分位数？

发布于 2025-01-10 03:11:03 字数 978 浏览 0 评论 0原文

我试图绘制一个变量相对于另一个变量的连续百分位数。我找到了 gamlss 包，它可以让我做很多事情。

data(iris)
(gamlss)
(gamm4)
r1<-gamlss(Sepal.Length~cs(Sepal.Width, 3), sigma.formula=~cs(Sepal.Width, 3), data=iris, family=BCT)

centiles(r1,cent = c(2.5, 5, 10, 50, 90, 95, 97.5), 
         pch=1,cex=0.5,col="black", 
         main="Centile curves for sepal width", 
         iris$Sepal.Width)

但是，我有多级数据，并且 id 喜欢解释这一点，因为该模型中的上限主要代表分布最高的组。

我找到了 gamm4 包，它似乎可以解释多级数据，但我不确定输出。

r2 <- gamm4(Sepal.Length~s(Sepal.Width),data=iris,random=~(1|Species), family= "poisson")
plot.gam(r2$gam)

我想知道是否有办法添加百分位数线，或者在多级 GAMLSS 中这是不可能的吗？另外，输出似乎只给出了关联线/曲线，是否可以显示绝对值？

提前致谢

原文

Im trying to plot continuous percentiles of a variable against another variable. Ive found the package gamlss which lets me do pretty much that.

data(iris)
(gamlss)
(gamm4)
r1<-gamlss(Sepal.Length~cs(Sepal.Width, 3), sigma.formula=~cs(Sepal.Width, 3), data=iris, family=BCT)

centiles(r1,cent = c(2.5, 5, 10, 50, 90, 95, 97.5), 
         pch=1,cex=0.5,col="black", 
         main="Centile curves for sepal width", 
         iris$Sepal.Width)

However, I have multilevel data, and id like to account for that, since the upper limit in this model primarily represents the group with the highest distribution.

Ive found the package gamm4, which seems to account for multilevel data, however im not sure about the output.

r2 <- gamm4(Sepal.Length~s(Sepal.Width),data=iris,random=~(1|Species), family= "poisson")
plot.gam(r2$gam)

I was wondering if there is a way to add percentile lines, or is that not possible in a multilevel GAMLSS?
Also, it seems the output only gives an association line/curve, is it possible to show absolute values?

Thanks in advance

分享到QQ

分享到微博