按R中的列汇总数据帧

Question

我有这个数据框 df:

df <- structure(list(App = structure(c(4L, 4L, 3L, 3L, 2L, 2L, 1L), .Label = c("DB", 
"End", "Mid", "Web"), class = "factor"), Server = structure(c(5L, 
6L, 1L, 2L, 3L, 4L, 7L), .Label = c("GServer101", "Hserver103", 
"JServer100", "Kserver200", "Server101", "Server102", "Xdb101"
), class = "factor"), Process1 = c(1L, 5L, 1L, 1L, 1L, 1L, 1L
), Process2 = c(1L, 1L, 1L, 4L, 1L, 1L, 1L), Process3 = c(NA, 
NA, NA, NA, NA, NA, NA), Process4 = c(NA, NA, NA, NA, NA, NA, 
NA), Process5 = c(NA, NA, NA, 1L, 1L, 1L, 1L)), .Names = c("App", 
"Server", "Process1", "Process2", "Process3", "Process4", "Process5"
), class = "data.frame", row.names = c(NA, -7L))

我希望能够按列总结 df 数据框以及计数和放置过程，如下所示。我需要知道每个应用程序按列名分组有多少进程。我将如何在 R 中执行此操作？

end <- structure(list(App = structure(c(4L, 3L, 2L, 1L), .Label = c("DB", 
"End", "Mid", "Web"), class = "factor"), Process1 = c(6L, 2L, 
2L, 1L), Process2 = c(2L, 5L, 2L, 1L), Process3 = c(0L, 0L, 0L, 
0L), Process4 = c(0L, 0L, 0L, 0L), Process5 = c(0L, 1L, 2L, 1L
)), .Names = c("App", "Process1", "Process2", "Process3", "Process4", 
"Process5"), class = "data.frame", row.names = c(NA, -4L))

Answer 1

您可以使用 dplyr:

library(dplyr)
df %>% 
      group_by(App) %>% 
      summarize_at(vars(starts_with("Process")), funs(sum(., na.rm=TRUE)))

# A tibble: 4 × 6
#     App Process1 Process2 Process3 Process4 Process5
#  <fctr>    <int>    <int>    <int>    <int>    <int>
#1     DB        1        1        0        0        1
#2    End        2        2        0        0        2
#3    Mid        2        5        0        0        1
#4    Web        6        2        0        0        0

或者如果列位置是首选，可以将位置传递给 .cols 参数：

df %>% 
       group_by(App) %>% 
       summarize_at(.cols=3:7, funs(sum(., na.rm=TRUE)))

# A tibble: 4 × 6
#     App Process1 Process2 Process3 Process4 Process5
#  <fctr>    <int>    <int>    <int>    <int>    <int>
#1     DB        1        1        0        0        1
#2    End        2        2        0        0        2
#3    Mid        2        5        0        0        1
#4    Web        6        2        0        0        0

Answer 2

这是一个使用data.table

的方法

library(data.table)
# convert df to data.table
setDT(df)

df[, lapply(.SD, sum, na.rm=TRUE), .SDcols=Process1:Process5, by="App"]
   App Process1 Process2 Process3 Process4 Process5
1: Web        6        2        0        0        0
2: Mid        2        5        0        0        1
3: End        2        2        0        0        2
4:  DB        1        1        0        0        1

或使用列位置代替列名

df[, lapply(.SD, sum, na.rm=TRUE), .SDcols=3:7, by="App"]
   App Process1 Process2 Process3 Process4 Process5
1: Web        6        2        0        0        0
2: Mid        2        5        0        0        1
3: End        2        2        0        0        2
4:  DB        1        1        0        0        1

如果这是新的，这里有一个快速分解。 lapply(.SD, sum, na.rm=TRUE) 表示 sum 所有列的 na.rm=TRUE，.SDcols=3:7 或 .SDcols=Process1:Process5 将此操作子集到所需的列，by=App 对操作进行分组.

按R中的列汇总数据帧

summarizing data frame by columns in R

r

summary

dataframe