使用diff()函数创建条形图，以确定下一个事件的生存变化

Question

我有一个生存数据集df_survival，它有四列Date、Tank、Feed_Group和Census。

下面是创建数据集的代码：

Date <- c("2020-10-15", "2020-10-15", "2020-10-15", "2020-10-15", "2020-10-15", "2020-10-15","2020-10-15", "2020-10-15","2020-12-05","2020-12-05","2020-12-05","2020-12-05","2020-12-05","2020-12-05","2020-12-05","2020-12-05","2021-01-29","2021-01-29","2021-01-29","2021-01-29","2021-01-29","2021-01-29","2021-01-29","2021-03-13","2021-03-13","2021-03-13","2021-03-13","2021-03-13","2021-03-13","2021-03-13","2021-03-13","2021-04-30","2021-04-30","2021-04-30","2021-04-30","2021-04-30","2021-04-30","2021-04-30","2021-04-30","2021-04-30", "2021-07-06", "2021-07-06", "2021-07-06", "2021-07-06", "2021-07-06", "2021-07-06", "2021-07-06", "2021-07-06")
Tank <- c(1,2,3,4,5,6,7,8,1,2,3,4,5,6,7,8,1,2,3,4,5,6,7,8,1,2,3,4,5,6,7,8,1,2,3,4,5,6,7,8,1,2,3,4,5,6,7,8)
Feed_Group <- c("Gemma","Gemma","Gemma","Gemma","Mysis","Mysis","Mysis","Mysis","Gemma","Gemma","Gemma","Gemma","Mysis","Mysis","Mysis","Mysis","Gemma","Gemma","Gemma","Gemma","Mysis","Mysis","Mysis","Mysis","Gemma","Gemma","Gemma","Gemma","Mysis","Mysis","Mysis","Mysis","Gemma","Gemma","Gemma","Gemma","Mysis","Mysis","Mysis","Mysis","Gemma","Gemma","Gemma","Gemma","Mysis","Mysis","Mysis","Mysis") 
Census <- c(20,20,20,20,20,20,20,20,15,17,14,10,14,20,16,16,13,16,12,10,14,20,16,16,13,15,11,10,14,20,16,16,13,15,10,9,14,20,15,16,13,14,10,8,14,19,15,15)


df_survival <- data.frame(Date,Tank,Feed_Group,Census)

我创建了这个可视化系统，它显示了随着时间的推移，每个群体的平均人口普查结果，并带有标准错误。

​

​

现在，我想要创建一个条形图，显示在一个事件到下一个事件中有多少动物死亡。因此，从事件1到事件2，引理组有6只动物减少，从事件2减少到事件3，1.25只动物减少。

我可以像下面这样使用diff()函数来创建一个tibble，它为我提供了一个与我想要的完全一样的Feed_Group和Census_Dead列。

df_survival %>%
  group_by(Feed_Group,Date) %>%
  summarise(Census = mean(Census)) %>%
  summarise(Census_Dead = diff(Census))

但是，我很可能会丢失Date列，因为第一个点没有0的差异，这是有道理的。不幸的是，这使得作图变得很困难。

我知道我可以创建data.frame并手动将这些数字与相应的日期放在一起，因为数据集太小了，但是我很好奇是否有可能操作当前的数据以获得如下内容：

​

Answer 1

可以使用dplyr::summarize为每个组获取更少的行(而不仅仅是聚合到一行)。或者，您可以使用dplyr::group_modify。

df_survival |>
  group_by(Date, Feed_Group) |>
  summarize(Census = mean(Census)) |>
  ungroup() |>
  arrange(Date) |>
  group_by(Feed_Group) |>
  summarize(Census_Dead = -diff(Census),
            Date = Date[-1]) |>
  ggplot(aes(y = Census_Dead, x= Date)) +
  geom_bar(aes(fill = Feed_Group), position = "dodge", stat = "identity")

​

​