从累积数据中获取反向发生率数据？

Question

我有一个数据框，其中包含日期数据和累计计数。我正在尝试对 cumsum 进行 反转以获取每日计数但还获取每组的计数。我正在尝试从数据框 A 转到数据框 B。我正在使用 R 和 tidyr.

这是代码：

df <- data.frame(cum_count = c(5, 14, 50, 5, 14, 50), state = c("Alabama", "Alabama", "Alabama", "NY", "NY", "NY"), Year = c(2012:2014, 2012:2014))

Dataframe A cum_count state Year 1 5 Alabama 2012 2 14 Alabama 2013 3 50 Alabama 2014 4 5 NY 2012 5 14 NY 2013 6 50 NY 2014

Dataframe B cum_count state Year 1 5 Alabama 2012 2 9 Alabama 2013 3 36 Alabama 2014 4 5 NY 2012 5 9 NY 2013 6 36 NY 2014

我试过使用 diff 函数:

df <- df %>%group_by(state)%>% mutate(daily_count = diff(cum_count))

但是我明白了

Error: Column daily_count must be length 3 (the number of rows) or one, not 2

告诉我你的想法。

谢谢！

Answer 1

diff returns length 比原来的长度少一并且 mutate 要求输出列与原来的 length 相同（或者长度 1，可回收）。我们可以附加一个值 NA 或 'cum_count'

的 first 值

library(dplyr)
df %>%
  group_by(state)%>%
  mutate(daily_count = c(first(cum_count), diff(cum_count)))
# A tibble: 6 x 4
# Groups:   state [2]
#  cum_count state    Year daily_count
#      <dbl> <fct>   <int>       <dbl>
#1         5 Alabama  2012           5
#2        14 Alabama  2013           9
#3        50 Alabama  2014          36
#4         5 NY       2012           5
#5        14 NY       2013           9
#6        50 NY       2014          36

或者为此目的，使用 lag 并从列本身中减去

df %>%
    group_by(state)%>%
    mutate(daily_count = replace_na(cum_count - lag(cum_count), first(cum_count)))

从累积数据中获取反向发生率数据？

Obtaining back incidence data from cumulative data?

diff

r

cumsum