Rowsums conditional on column name

问题

My data.frage looks like this:

   VAR1 VAR2 AUS1 AUS2 AUS3 AUS4 ... AUS56 VAR3 VAR4
   A    D    23   234  34   856  ... 99    0    FCK
   B    D    55   76   55   36   ... 6456  0    XYC

I'd like R to add a new variable AUS which shows the rowsums of the variables AUS1 to AUS56, preferably with dplyr. AUS1 to AUS56 can then be deleted.

回答1:

You can try use rowSums in combination with grep:

df %>% mutate(AUS_sum = rowSums(.[grep("AUS", names(.))]))

回答2:

Here is another option using tidyverse syntax

library(tidyverse)
df1 %>% 
     select(matches("AUS")) %>% 
     reduce(`+`) %>%
     mutate(df1, AUS_sum = .)
#   VAR1 VAR2 AUS1 AUS2 AUS3 AUS4 AUS56 VAR3 VAR4 AUS_sum
#1    A    D   23  234   34  856    99    0  FCK    1246
#2    B    D   55   76   55   36  6456    0  XYC    6678

With the devel version of dplyr (soon to be released 0.6.0) we can create a function with quosures and make it more dynamic. Here, the enquo does similar functionality as substitute from base R by taking the input arguments and converting it to quosure, with quo_name, we convert it to string where matches takes string argument. The lhs name can also be created as string ('newN') and within the mutate/summarise/group_by, we unquote (!! or UQ) to evaluate the string

fSum <- function(dat, pat){
  pat <- quo_name(enquo(pat))
  newN <- paste0(pat, "_sum")
  newSum <- dat %>%
            select(matches(pat)) %>%
            reduce(`+`)
  dat %>%
      mutate(!!newN :=  newSum)
}

fSum(df1, AUS)
#    VAR1 VAR2 AUS1 AUS2 AUS3 AUS4 AUS56 VAR3 VAR4 AUS_sum
#1    A    D   23  234   34  856    99    0  FCK    1246
#2    B    D   55   76   55   36  6456    0  XYC    6678

Based on the OP's comment on the other post about removing the columns that used for sum, we can modify the function

fSumN <- function(dat, pat){
  pat <- quo_name(enquo(pat))
  newN <- paste0(pat, "_sum")
  newSum <- dat %>%
            select(matches(pat)) %>%
            reduce(`+`)
  dat %>%
       select(-matches(pat)) %>%
       mutate(!!newN :=  newSum)
}

fSumN(df1, AUS)
#     VAR1 VAR2 VAR3 VAR4 AUS_sum
#1    A    D    0  FCK    1246
#2    B    D    0  XYC    6678

data

df1 <- structure(list(VAR1 = c("A", "B"), VAR2 = c("D", "D"), AUS1 = c(23L, 
55L), AUS2 = c(234L, 76L), AUS3 = c(34L, 55L), AUS4 = c(856L, 
36L), AUS56 = c(99L, 6456L), VAR3 = c(0L, 0L), VAR4 = c("FCK", 
"XYC")), .Names = c("VAR1", "VAR2", "AUS1", "AUS2", "AUS3", "AUS4", 
 "AUS56", "VAR3", "VAR4"), class = "data.frame", row.names = c(NA, 
-2L))

回答3:

In base R:

df$AUS <- rowSums(df[,grep('AUS', names(df))])

来源：https://stackoverflow.com/questions/43515925/rowsums-conditional-on-column-name

标签

dplyr

rowsum