ggplot 可视化单个值的置信区间（准确率）

Question

如果有人能告诉我如何不仅绘制每个分类器的准确性，还绘制置信区间（AccuracyLower、AccuracyUpper），我会很高兴。每当我使用 ggplot 时，绘图只会显示准确性，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")

Answer 1

这是您要找的东西吗：

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

作为旁注，您的绘图也工作得很好 - 它绘制了所有点，但它们非常接近所以它们重叠（如果比例从 0 到 1）

ggplot 可视化单个值的置信区间（准确率）

ggplot visualize confidence interval of single values (accuracy rates)

plot

r

ggplot2

confidence-interval