ggplot可视化单个值的置信区间准确率

问题描述

如果有人可以向我展示如何不仅可以绘制每个分类器的精度，还可以绘制置信区间（AccuracyLower，AccuracyUpper），我将非常喜欢。每当我使用ggplot时，绘图只会可视化Accuracy，而不是其他值。我的数据框如下所示：

compmods <- data.frame(Model = c("SVM","XGB","Random forest","Neural net"),Accuracy = c(0.7324477,0.7460202,0.7634408,0.7420827),AccuracyLower = c(0.7297001,0.7433180,0.7608018,0.7393670),AccuracyUpper = c(0.7351819,0.7487083,0.7660646,0.7447845)) 
#this does not produce the desired plot: 
compmods %>%
  ggplot(aes(x = Model,y = Accuracy)) + theme_bw() +
  geom_point(aes(Model,AccuracyLower),shape = 20) +
  geom_point(aes(Model,AccuracyUpper))+
  geom_point() +
  ylim(0,1) +
  geom_hline(yintercept = 0.695628,color = "Red") +
  labs(title = "Average classifier accuracy compared to no information rate")

解决方法

这是您要寻找的东西吗

compmods %>% 
ggplot(aes(Model)) + 
geom_errorbar(aes(ymin = AccuracyLower,ymax = AccuracyUpper)) + 
geom_point(aes(y = Accuracy),color = "red",size = 5)

请注意，您的绘图也可以正常工作-可以绘制所有点，但是它们非常接近，因此它们会重叠（如果比例从0到1）