4

有没有办法直接从 MuMIn model.avg() 为具有置信带的不同变量绘制模型平均汇总输出。以前我一直在使用 ggplot 和 ggpredict 从实际模型中绘制项,但我一直无法找到一种方法来绘制平均模型的结果。

显然,我可以手动绘制斜率和截距,但是获得准确的置信带并从 confint() 绘制并不理想,而且我还没有从看起来正确的区间中获得置信带。

library(MuMIn)
#Dummy Data
a <- seq(1:5)
set.seed(1)
b <- sample(1:100,5)
c <- sample(1:100,5)
d <-sample(1:100,5)
df <- data.frame(a,b,c,d)

Dredged <- dredge(lm(a ~ b + c + d, data=df), rank=AIC)
ModelAvg <- model.avg(Dredged, subset=delta<=2)


CI <- confint(ModelAvg, full=T) # get confidence intervals
summary(ModelAvg)


#I want to be able to create a graph for each term from the averaged output with its estimate, SE, and Confidence bands


#Output - I've only left the relevant part of the output, my actual data ends up with 5 component models
#Call:
#model.avg(object = Dredged, subset = delta <= 2)
#Component models: 
#    df logLik   AIC delta weight
#12   4  -1.32 10.63  0.00   0.69
#123  5  -1.10 12.21  1.58   0.31

#Model-averaged coefficients:  
#(full average) 
#             Estimate Std. Error Adjusted SE z value Pr(>|z|)
#(Intercept)  4.933497   1.308953    7.725454   0.639    0.523
#b            0.021946   0.010320    0.048539   0.452    0.651
#c           -0.044848   0.012076    0.067954   0.660    0.509
#d           -0.002275   0.014081    0.088694   0.026    0.980


4

1 回答 1

3

我不太确定我是否理解您为什么要质疑“confint()”输出,并且其输出的有效性确实是与绘图问题不同的问题。

要绘制系数 +/- SE,请调整。SE 和 95% CI,请尝试以下操作。这使用完整的模型平均值,因为您full=T在 CI 中使用了参数。

该图表不是最漂亮的,但它确实可以完成工作 - 如果您想要一个更好的图表,请告诉我。我没有绘制截距图,因为在这种情况下,估计值远大于系数,但所有数据都采用易于绘制的格式。

library(MuMIn)
#Dummy Data
a <- seq(1:5)
set.seed(1)
b <- sample(1:100,5)
c <- sample(1:100,5)
d <-sample(1:100,5)
df <- data.frame(a,b,c,d)

options(na.action = "na.fail") # needed for dredge to work
Dredged <- dredge(lm(a ~ b + c + d, data=df), rank=AIC)
ModelAvg <- model.avg(Dredged)
mA<-summary(ModelAvg) #pulling out model averages
df1<-as.data.frame(mA$coefmat.full) #selecting full model coefficient averages

CI <- as.data.frame(confint(ModelAvg, full=T)) # get confidence intervals for full model
df1$CI.min <-CI$`2.5 %` #pulling out CIs and putting into same df as coefficient estimates
df1$CI.max <-CI$`97.5 %`# order of coeffients same in both, so no mixups; but should check anyway
setDT(df1, keep.rownames = "coefficient") #put rownames into column
names(df1) <- gsub(" ", "", names(df1)) # remove spaces from column headers

绘制所有三个误差线(SE、adj. SE、95% CI)

ggplot(data=df1[2:4,], aes(x=coefficient, y=Estimate))+ #excluding intercept because estimates so much larger
  geom_point(size=10)+ #points for coefficient estimates
  theme_classic(base_size = 20)+ #clean graph
  geom_errorbar(aes(ymin=Estimate-Std.Error, ymax=Estimate+Std.Error), colour ="red", # SE
             width=.2, lwd=3) +
  geom_errorbar(aes(ymin=Estimate-AdjustedSE, ymax=Estimate+AdjustedSE), colour="blue", #adj SE
              width=.2, lwd=2) +
  geom_errorbar(aes(ymin=CI.min, ymax=CI.max), colour="pink", # CIs
                width=.2,lwd=1) 

这会产生下图。红色是 SE,蓝色是 adj。SE 和粉红色是 95% CI。 在此处输入图像描述

用更好的图表编辑:

ggplot(data=df1[2:4,], aes(x=coefficient, y=Estimate))+ #again, excluding intercept because estimates so much larger
      geom_hline(yintercept=0, color = "red",linetype="dashed", lwd=1.5)+ #add dashed line at zero
      geom_errorbar(aes(ymin=Estimate-AdjustedSE, ymax=Estimate+AdjustedSE), colour="blue", #adj SE
                  width=0, lwd=1.5) +
      coord_flip()+ # flipping x and y axes
      geom_point(size=8)+theme_classic(base_size = 20)+ ylab("Coefficient")

在此处输入图像描述

于 2019-11-30T09:57:44.537 回答