减少名称与模式匹配的列

Question

我正在尝试在我的数据表中创建多个列，这些列表示按行匹配给定模式的列的总和。

df <- data.frame(first_column = c("Alpha", "Beta", "Charlie", "Tango", "Alpha, Beta,Alpha", "Alpha,Beta,Charlie", 'Tango,Tango,Tango,Tango', 'Tango,Tango,Tango, Tango', 'Tango,Tango,Tango, Tango , Alpha,Beta,Charlie, Alpha, Alpha ,Alpha '),
                 number_1 = 1:9,
                 number_2 = 11:19,
                 number_3 = 2:10,
                 number_4 = 12:20) 

testing <- df %>%
  mutate(number_1 = as.numeric(number_1),
         number_2 = as.numeric(number_2))%>%
  as.data.table 


testing3 = testing[,`:=` ("Total 1" = Reduce(`+`, grep("number_1|number_2", names(testing), value = TRUE)),
                          "Total 2" = Reduce(`+`, grep("number_3|number_4", names(testing), value = TRUE)))]

这是我尝试过的，但没有成功。我应该看到的是 Total 1 row 1, 12 和 Total 2 row 1, 14

但是我做不到

Answer 1

原因是 grep returns 只有带 value = TRUE 的列名，我们需要列的值，使用 .SD 对列进行子集化列名

library(data.table)
testing[,`:=` (
   "Total 1" = Reduce(`+`, .SD[, grep("number_1|number_2", names(.SD),
              value = TRUE), with = FALSE]),
    "Total 2" = Reduce(`+`, .SD[, grep("number_3|number_4", names(.SD), 
     value = TRUE), with = FALSE]))]

-输出

> testing
                                                          first_column number_1 number_2 number_3 number_4 Total 1 Total 2
                                                                <char>    <num>    <num>    <int>    <int>   <num>   <int>
1:                                                               Alpha        1       11        2       12      12      14
2:                                                                Beta        2       12        3       13      14      16
3:                                                             Charlie        3       13        4       14      16      18
4:                                                               Tango        4       14        5       15      18      20
5:                                                   Alpha, Beta,Alpha        5       15        6       16      20      22
6:                                                  Alpha,Beta,Charlie        6       16        7       17      22      24
7:                                             Tango,Tango,Tango,Tango        7       17        8       18      24      26
8:                                            Tango,Tango,Tango, Tango        8       18        9       19      26      28
9: Tango,Tango,Tango, Tango , Alpha,Beta,Charlie, Alpha, Alpha ,Alpha         9       19       10       20      28      30

如果有多个集合，我们还可以创建一个命名的 list , Filter list 元素基于名字的出现

lst_names <- list(c("number_1", "number_2"), 
                 c("number_3", "number_4"), 
                 c("number_5", "number_6"))
names(lst_names) <- paste("Total", seq_along(lst_names))
lst_names_sub <- Filter(length, lapply(lst_names, function(x)
        intersect(x, names(testing))))
testing[, names(lst_names_sub) := lapply(lst_names_sub, function(x) 
         Reduce(`+`, .SD[, x, with = FALSE]))]

减少名称与模式匹配的列

Reduce columns that whose names matches a pattern

datatable

r