ggplot 中的 Geom_errorbar 未显示正确的标准错误

Question

我想在我的 geom_col ggplot 中添加误差线。误差线确实添加到我的情节中，但误差线不正确。有人可以帮我弄清楚如何调整我的代码吗？我认为y值是错误的。它可能不应该是长度，而是我的数据框中“数字”的实际值。 如果这是一个非常简单的问题，我深表歉意，我在 R 中很新。 提前谢谢你。

number_of_sprout <- data.frame(
  Condition = c("RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl"), 
  Number = c(8, 11, 11, 17, 18, 13, 16, 4, 11, 9, 12, 11, 15, 11, 17, 8, 18, 9, 7, 11, 11, 12, 8, 12, 6, 8, 10, 19, 9, 14, 7, 9, 12, 11, 8, 12, 6, 8, 10, 8, 11, 11, 10, 7, 6, 6, 6, 9, 11, 10, 11, 12, 8, 14, 7, 8, 11, 9, 9, 12, 10, 7, 13, 17, 15, 9, 14, 11, 11, 8, 14, 13, 7, 8, 6, 7, 9, 7, 10, 7), 
  timepoint = c(24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48)
)


sph1 <- ggplot(data = number_of_sprout, aes(x = factor(timepoint, ordered = TRUE), y=Number, fill = Condition)) +
  stat_summary(fun = mean, geom = "col", position = position_dodge()) +
  scale_fill_manual(values = c("slategray4", "thistle3")) +
  geom_point(position = position_dodge(width = 0.9)) +
    labs(x = "Hours", y = "Number of sprouts") +
    theme_classic() +
    ggtitle("Spheroids - Donor 37") +
    theme_bw() +
    theme(text=element_text(family="Times New Roman", face="bold")) +
    theme(aspect.ratio = 1.5/1)

se5 <- function(y) sd(y)/length(y)   # to calculate standard error in the mean
sph1+stat_summary(geom="errorbar",position=position_dodge(width=0.9),
                 fun.data=function(y)c(ymin=mean(y)-se5(y),ymax=mean(y)+se5(y)), width=0.2)

Answer 1

作为不同的答案 - 不同的主题。您是否考虑过对配对数据进行不同形式的可视化？我会将其显示为关系散点图，第一个度量在 x 轴上，第二个在 y 轴上。

如您所见，通过这种方式，两个测量值之间的关系更清晰可见，并且您还可以比条形图更好地看到差异（或本例中的相似性）。

library(tidyverse)
number_of_sprout <- data.frame(
 Condition = c("RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "RBP7 KO", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl", "Ctrl"), 
 Number = c(8, 11, 11, 17, 18, 13, 16, 4, 11, 9, 12, 11, 15, 11, 17, 8, 18, 9, 7, 11, 11, 12, 8, 12, 6, 8, 10, 19, 9, 14, 7, 9, 12, 11, 8, 12, 6, 8, 10, 8, 11, 11, 10, 7, 6, 6, 6, 9, 11, 10, 11, 12, 8, 14, 7, 8, 11, 9, 9, 12, 10, 7, 13, 17, 15, 9, 14, 11, 11, 8, 14, 13, 7, 8, 6, 7, 9, 7, 10, 7), 
 timepoint = c(24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48)
)
number_of_sprout %>%
 pivot_wider(names_from = timepoint, values_from = Number, 
             names_prefix = "t", values_fn = list) %>%
 unnest(c(t24, t48)) %>%
ggplot(aes(t24, t48)) +
 geom_abline(intercept = 0, slope = 1, color = "grey") +
 geom_point(aes(color = Condition)) +
 scale_color_brewer(palette = "Set1", direction = -1) +
 coord_equal(xlim = c(0,NA), ylim = c(0, NA)) +
 cowplot::theme_cowplot()

^{由reprex package (v2.0.0) 于 2021 年 4 月 29 日创建}

Answer 2

我假设您只是对平均值的标准误有错误的公式。我已经将它保存在我的个人实用程序包中，所以我不需要查找它 - 在某些时候厚颜无耻地从 this thread 偷走了。

另一个注意事项，也许检查http://data-to-viz.com/caveat/error_bar.html 更一般的警告为什么误差线可能不理想（如果你绘制整个数据，一般来说不是很有必要）

# replace your function
se5 <- function(x, na.rm=FALSE) {
  if (na.rm) x <- na.omit(x)
  sqrt(var(x)/length(x))
}  # to calculate standard error in the mean
sph1+stat_summary(geom="errorbar",position=position_dodge(width=0.9),
                  fun.data=function(y)c(ymin=mean(y)-se5(y),ymax=mean(y)+se5(y)), width=0.2)