从R中的各个单元格中提取摘要数据

Question

Here is the Data:

data <-data.frame(
    "start"= c("go",NA,NA,NA,"go",NA,"go"),
    "number"= c(31,32,1,29,61,17,72),
    "info"= c("c","k","s","u","b","i","n"))

   start number info
1    go     31    c
2  <NA>     32    k
3  <NA>      1    s
4  <NA>     29    u
5    go     61    b
6  <NA>     17    i
7    go     72    n

And I want to produce a summary table that prints the info on each line where Start = "go"

However, I want the Number column to be summed from all the cells after a "go" until the next "go" so that the results look as follows:

final <- data.frame(
"start"=c("go","go","go"),
"number"=c(93,78,72),
"info"=c("c","b","n"))

   start number info
1    go     93    c
2    go     78    b
3    go     72    n

Thanks for your help.

Answer 1

A strategy in base R is to perform the subsetting and perform the summation in separate operations and then merge the results together. Here, we can use cbind for the merge, since the two datasets were constructed to line up.

cbind(data[!is.na(data$start), c(1, 3)],
      sum=aggregate(number ~ cumsum(!is.na(start)), data=data, sum)[,2])
  start info sum
1    go    c  93
5    go    b  78
7    go    n  72

I use !is.na to select the proper rows, which works in this example. If you have other, non-NA values you want excluded, you can use !is.na(data$start) & data$start == "go" . aggregate is used to perform the summation and the grouping in the second argument uses the same method, but performs a cumulative sum on the result.

Answer 2

You could use dplyr:

data <-data.frame(
  start= c("go",NA,NA,NA,"go",NA,"go"),
  number= c(31,32,1,29,61,17,72),
  info= c("c","k","s","u","b","i","n"),stringsAsFactors = F)

library(dplyr)
data$group = cumsum(!is.na(data$start))
data %>% group_by(group) %>% summarize(n=sum(number), info=info[1])

Output

  group     n  info
1     1    93     c
2     2    78     b
3     3    72     n

Optionally you could add

 %>% mutate(start="go") %>% select(-group)

to get to your requested output, but I am not sure if that actually adds value. Hope this helps!

Answer 3

Here is an option using data.table

library(data.table)
setDT(data)[, .(start = start[!is.na(start)], n = sum(number), 
     info = info[1]), .(grp = cumsum(!is.na(start)))][, grp := NULL][]
#   start  n info
#1:    go 93    c
#2:    go 78    b
#3:    go 72    n