如何按行取选择列的平均值

Question

我的tibble 插入到最后。

我有一个包含不同类型列的数据框（它们是不同的重复）。前四列应保持原样； start_with()（我这样写是因为函数可能有用）“树冠”应该概括为mean，以及“understory”（写成“under”）和“tree diameter” -at-breast-height" (tdbh) – 包括na.rm = TRUE。我怎样才能做到这一点？我该如何总结这样的列？

数据（头部）：

structure(list(Site = c("Bala", "Bala", "Bala", "Bala", "Bala", 
"Bala"), Classification = c("Primary forest", "Primary forest", 
"Primary forest", "Primary forest", "Primary forest", "Primary forest"
), transect = c(1, 1, 1, 1, 1, 2), point = c(1, 2, 3, 4, 5, 1
), canopy1 = c(10, 2, 5, 10, 2, 4), canopy2 = c(4, 3, 2, 3, 2, 
6), canopy3 = c(5, 2, 3, 4, 5, 1), canopy4 = c(5, 3, 2, 2, 2, 
2), canopy5 = c(3, 9, 6, 7, 4, 8), under1 = c(15, 17, 4, 23, 
9, 27), under2 = c(13, 1, 0, 0, 0, 0), under3 = c(1, 4, 4, 4, 
5, 23), under4 = c(19, 9, 0, 5, 0, 0), Numtrees = c(4, 3, 6, 
1, 3, 7), t1dbh = c(110, 100, 50, 60, 100, 70), t2dbh = c(80, 
60, 60, NA, 70, 80), t3dbh = c(80, 110, 70, NA, 100, 50), t4dbh = c(90, 
NA, 110, NA, NA, 60), t5dbh = c(NA, NA, 90, NA, NA, 50), t6dbh = c(NA, 
NA, 110, NA, NA, 60), t7dbh = c(NA, NA, NA, NA, NA, 250), t8dbh = c(NA_real_, 
NA_real_, NA_real_, NA_real_, NA_real_, NA_real_), t9dbh = c(NA_real_, 
NA_real_, NA_real_, NA_real_, NA_real_, NA_real_), t10dbh = c(NA_real_, 
NA_real_, NA_real_, NA_real_, NA_real_, NA_real_)), row.names = c(NA, 
-6L), class = c("tbl_df", "tbl", "data.frame"))

所需的输出以：

开头

# A tibble: 2 x 7
  Site  Classification transect point canopy understory tdbh 
  <chr> <chr>             <dbl> <dbl>  <dbl> <chr>      <chr>
1 Bala  Primary forest        1     1    5.4 ...        ...  
2 Bala  Primary forest        1     2    3.8 ...        ...  

我希望这可能只使用基本 R 和 tidyverse 中的任何内容（可能是 dplyr 和/或 tidyr）...

编辑：我知道mutate(canopy = mean(c(canopy1, canopy2, ...))) 应该可以正常工作，但这有两个问题：首先，它添加了一个列而不是替换。这很麻烦，但并不可怕。但是，我必须列出所有内容。这是低效答案的标志。

Answer 1

使用sapply。

cbind(df[1:4], sapply(c("canopy", "under", "dbh"), function(x) 
  rowMeans(df[grep(x, names(df))], na.rm=TRUE)))
#   Site Classification transect point canopy under      dbh
# 1 Bala Primary forest        1     1    5.4 12.00 90.00000
# 2 Bala Primary forest        1     2    3.8  7.75 90.00000
# 3 Bala Primary forest        1     3    3.6  2.00 81.66667
# 4 Bala Primary forest        1     4    5.2  8.00 60.00000
# 5 Bala Primary forest        1     5    3.0  3.50 90.00000
# 6 Bala Primary forest        2     1    4.2 12.50 88.57143

Answer 2

你不能这样做：

df$canopy <- rowMeans(df[grep("^canopy", names(df))])

或者，在 tidyverse-speak 中（并使用 select 删除中间列以显示结果）：

 df %>% 
   mutate(canopy = rowMeans(select(., starts_with("canopy")))) %>%
   select(-(5:24))

#> # A tibble: 6 x 5
#>   Site  Classification transect point canopy
#>  <chr> <chr>             <dbl> <dbl>  <dbl>
#> 1 Bala  Primary forest        1     1    5.4
#> 2 Bala  Primary forest        1     2    3.8
#> 3 Bala  Primary forest        1     3    3.6
#> 4 Bala  Primary forest        1     4    5.2
#> 5 Bala  Primary forest        1     5    3  
#> 6 Bala  Primary forest        2     1    4.2

Answer 3

使用tidyverse 包的解决方案。我们可以用目标字符串创建一个向量，然后使用map_dfc 和mutate 动态计算平均值。之后，我们可以将计算得到的列合并到原始数据框中。

library(tidyverse)

# Set the target column names
target <- c("canopy", "under", "dbh")

# 
dat2 <- map_dfc(target, function(x){
  temp <- dat %>%
    mutate("{x}" := rowMeans(select(., contains(x)), na.rm = TRUE), .keep = "none")
})

dat3 <- dat %>% 
  select(-contains(target)) %>%
  bind_cols(dat2)

print(dat3)
# # A tibble: 6 x 8
#   Site  Classification transect point Numtrees canopy under   dbh
#   <chr> <chr>             <dbl> <dbl>    <dbl>  <dbl> <dbl> <dbl>
# 1 Bala  Primary forest        1     1        4    5.4 12     90  
# 2 Bala  Primary forest        1     2        3    3.8  7.75  90  
# 3 Bala  Primary forest        1     3        6    3.6  2     81.7
# 4 Bala  Primary forest        1     4        1    5.2  8     60  
# 5 Bala  Primary forest        1     5        3    3    3.5   90  
# 6 Bala  Primary forest        2     1        7    4.2 12.5   88.6