R 将行汇总为一行（连续变量和因子变量）

Question

我正在尝试将一天中的一堆行汇总为一行。如果可能的话，我希望它在 dplyr 中。我知道我的代码远非正确，但这就是我得到的结果：

data %>%
  group_by(DAY) %>%
  summarise_each(funs(Sum = n()), SEX, GROUP, TOTAL)

原文：

DAY SEX GROUP   TOTAL       
7/1/14  FEMALE  A   1       
7/1/14  FEMALE  B   1       
7/1/14  FEMALE  B   1       
7/1/14  FEMALE  A   1       
7/1/14  MALE    A   1       
7/1/14  MALE    B   2

新：

DAY     FEMALE  MALE    GROUP_A GROUP_B TOTAL
7/1/14  4       2       3       3       7

Answer 1

您计算总计（总和）和其他列 (table) 的方式大不相同，因此您可能必须分别执行这些步骤。计算总数很容易。对于表格，我建议使用 tidyr 如下：

# required packages
require(dplyr)
require(tidyr)

# calculations
data %>%
  group_by(DAY) %>%                     # group by day
  mutate(TOTAL = sum(TOTAL)) %>%        # first calculate total
  gather(key, value, -DAY, -TOTAL) %>%  # collapse
  unite(group, key, value) %>%          # get sensible column names
  group_by(DAY, TOTAL) %>%              # group by day and total
  do(as.data.frame(table(.$group))) %>% # table
  spread(Var1, Freq)                    # spread out

##      DAY TOTAL GROUP_A GROUP_B SEX_FEMALE SEX_MALE
## 1 7/1/14     7       3       3          4        2

Answer 2

这似乎有点神秘，但这是一个简短的咒语

dat %>% group_by(DAY) %>%
  summarise_each(funs(ifelse(is.numeric(.), sum(.), list(table(.))))) -> res

data.frame(DAY=res$DAY, t(unlist(res[, 2:ncol(res)])))
#      DAY SEX.FEMALE SEX.MALE GROUP.A GROUP.B TOTAL
# 1 7/1/14          4        2       3       3     7

在这里，如果不是数字，您只需将每一列汇总为 table，如果是，则对其求和（对于总计列）。这需要作为列表返回，因为 summarise_each 需要一个值。然后，结果扩展为常规 data.frame.

Answer 3

一种可能的方法：

library(reshape2)
library(data.table)

cbind(dcast(df, DAY~SEX), 
      dcast(df, DAY~GROUP)[-1], 
      setDT(df)[,.(total=sum(TOTAL)),DAY][,-1,with=F])

#     DAY FEMALE MALE A B total
#1 7/1/14      4    2 3 3     7

Answer 4

data.table 的另一种方式，在 data.frame 上测试超过一天。

require(data.table)
setDT(data)[, as.list(c(table(SEX), table(GROUP), TOTAL=sum(TOTAL))), by=DAY]

#      DAY FEMALE MALE A B TOTAL
#1: 7/1/14      3    0 1 2     3
#2: 8/1/14      1    2 2 1     4

编辑：另一个更少手动的选项（您不需要知道哪些变量是因子，哪些是数字），感谢@jangorecki 和@的一些帮助戴维阿伦伯格

wh_num <- sapply(data, is.numeric)[-1]
wh_fact <-sapply(data, is.factor)[-1]
setDT(data)[, as.list(c(lapply(.SD[, wh_fact, with = FALSE], table), 
                        lapply(.SD[, wh_num, with = FALSE], sum), 
                        recursive = TRUE)), by = DAY]

#      DAY SEX.FEMALE SEX.MALE GROUP.A GROUP.B TOTAL
#1: 7/1/14          3        0       1       2     3
#2: 8/1/14          1        2       2       1     4

数据

data <- structure(list(DAY = c("7/1/14", "7/1/14", "7/1/14", "8/1/14", 
"8/1/14", "8/1/14"), SEX = structure(c(1L, 1L, 1L, 1L, 2L, 2L
), .Label = c("FEMALE", "MALE"), class = "factor"), GROUP = structure(c(1L, 
2L, 2L, 1L, 1L, 2L), .Label = c("A", "B"), class = "factor"), 
    TOTAL = c(1L, 1L, 1L, 1L, 1L, 2L)), .Names = c("DAY", "SEX", 
"GROUP", "TOTAL"), row.names = c(NA, -6L), class = "data.frame")

R 将行汇总为一行（连续变量和因子变量）

R rolling up rows to a single row (continuous & factor variables)

r

dplyr