ggplot：记录 y 轴时添加回归线和 CI

Question

这是我的数据：

x

是输入值，

y

是使用

lm()

模型的预测值，

upr

和

lwr

是根据

lm()

模型计算的95%置信区间

library(tidyverse)
library(scales)
library(cowplot)
library(ggpmisc)

    fit_line <-dput(fit_line)
structure(list(x = c(0.048, 0.55, 0.44, 0.052, 0.029, 0.09, 0.2, 
0.48, 0.51, 0.11, 1.44, 0.15, 0.4, 1.12, 0.54), y = c(535.247309812533, 
1542.55243882316, 1321.82820736665, 543.273645501861, 497.122215288227, 
619.523834550474, 840.248066006986, 1402.09156425993, 1462.28908192988, 
659.655512997113, 3328.41212969858, 739.91886989039, 1241.56485047337, 
2686.30527455236, 1522.48659959984), lwr = c(-474.582987840204, 
754.520653106387, 576.903888431076, -461.460391144338, -537.324262756609, 
-338.409577605306, -2.1782812716142, 647.785886082738, 696.130234140302, 
-274.914182245542, 1249.20869628597, -150.884187631952, 498.439060263022, 
1155.6674836111, 740.548147246478), upr = c(1545.07760746527, 
2330.58422453993, 2066.75252630222, 1548.00768214806, 1531.56869333306, 
1577.45724670625, 1682.67441328559, 2156.39724243711, 2228.44792971946, 
1594.22520823977, 5407.61556311119, 1630.72192741273, 1984.69064068372, 
4216.94306549362, 2304.4250519532)), class = "data.frame", row.names = c("1", 
"2", "3", "4", "5", "6", "7", "8", "9", "10", "11", "12", "13", 
"14", "15"))

所以我尝试绘制的是与 CI 一致的，按 log y 比例。但是发生了： 所以首先当我不记录比例时：

ggplot() +
 # scale_y_log10(breaks = c(10, 100, 1000, 10000),labels = c("10", "100", "1000", "10000")  ) +
  coord_cartesian(xlim = c(0, 1.5),ylim = c(10, 10000)) +
  theme_cowplot() +
  annotation_logticks(sides = "l") +
  geom_ribbon(data = fit_line, aes(x,y,ymin = lwr, ymax = upr), fill = "grey80", alpha = 0.4) 

我不明白那个黑色矩形是从哪里来的？ 然后当我记录 y 尺度时，我丢失了一些 CI：

ggplot() +
  scale_y_log10(breaks = c(10, 100, 1000, 10000),labels = c("10", "100", "1000", "10000")  ) +
  coord_cartesian(xlim = c(0, 1.5),ylim = c(10, 10000)) +
  theme_cowplot() +
  annotation_logticks(sides = "l") +
  geom_ribbon(data = fit_line, aes(x,y,ymin = lwr, ymax = upr), fill = "grey80", alpha = 0.4) 

您能帮我找出代码中的问题吗？或者发生了什么？

Answer 1

annotation_logticks

函数从原始y轴值中获取log10，ggplot在使用

scale_y_log10

对轴进行log10变换后保存该值。如果您实际上不进行转换，则 logtickes 的值将非常小（在您的情况下为 1 到 5）。这就是为什么您会看到由压缩的逻辑标记组成的黑色矩形。

正如我在评论中提到的，对于第二个图，您会丢失一些间隔，因为

lwr

中有负值。您需要决定如何处理它们。您可能想将负值转换为 -Inf。

geom_ribbon(data = fit_line %>% mutate(lwr = ifelse(lwr < 0, -Inf, lwr)), 
            aes(x,y,ymin = lwr, ymax = upr), fill = "grey80", alpha = 0.4)