我们可以使用

library(dplyr)
library(tidyr)
library(data.table)
library(stringr)
df1 %>%
   pivot_longer(cols = where(is.numeric), names_to = 'year') %>% 
   as.data.table %>%
   cube( .(Mean = mean(value), SD = sd(value)),
     by = c("Fruit", "year")) %>% 
   filter(!if_all(Fruit:year, is.na)) %>% 
   unite(Fruit, Fruit, year, sep = "_", na.rm = TRUE) %>% 
   filter(str_detect(Fruit, "_|\\d+")) %>% 
   data.table::transpose(make.names = "Fruit", keep.names = "Measure")

• 输出

Measure Apples_2021 Apples_2022 Bananas_2021 Bananas_2022 Oranges_2021 Oranges_2022     2021     2022
1:    Mean    28.00000    42.00000    17.500000     54.50000    24.500000    77.500000 23.33333 58.00000
2:      SD    22.62742    18.38478     9.192388     33.23402     7.778175     6.363961 12.42041 23.57965

或者如果我们需要重复的列名

df1 %>%
    pivot_longer(cols = where(is.numeric), names_to = 'year') %>% 
   as.data.table %>%
   cube( .(Mean = mean(value), SD = sd(value)), by = c("Fruit", "year")) %>%
    mutate(Fruit = coalesce(Fruit, year)) %>%
   drop_na(year) %>%
   arrange(year, str_detect(Fruit, '\\d{4}', negate = TRUE)) %>% 
   select(-year) %>% 
   data.table::transpose(make.names = "Fruit", keep.names = "Measure")

• 输出

Measure     2021   Apples   Bananas   Oranges     2022   Apples  Bananas   Oranges
1:    Mean 23.33333 28.00000 17.500000 24.500000 58.00000 42.00000 54.50000 77.500000
2:      SD 12.42041 22.62742  9.192388  7.778175 23.57965 18.38478 33.23402  6.363961

数据

df1 <- structure(list(Fruit = c("Apples", "Bananas", "Apples", "Oranges", 
"Oranges", "Bananas"), `2021` = c(12L, 11L, 44L, 30L, 19L, 24L
), `2022` = c(29L, 31L, 55L, 73L, 82L, 78L)),
 class = "data.frame", row.names = c(NA, 
-6L))

我建议这可能是更好的（从长远来看）在一个长的格式，这一总结可以开始。这只是'平均'，并不难重复sd和合并这一点：

fruits <- c(NA, "Apples", "Oranges", "Bananas")
lapply(quux[,-1], function(yr) stack(sapply(fruits, function(z) mean(yr[is.na(z) | quux$Fruit %in% z])))) |>
  dplyr::bind_rows(.id = "year")
#   year   values     ind
# 1 2021 23.33333    <NA>
# 2 2021 28.00000  Apples
# 3 2021 24.50000 Oranges
# 4 2021 17.50000 Bananas
# 5 2022 58.00000    <NA>
# 6 2022 42.00000  Apples
# 7 2022 77.50000 Oranges
# 8 2022 54.50000 Bananas

其中ind中的NA表示所有果实，否则标记单个果实。

如果您将数据放入长格式，则可以使用聚合函数：

a <- aggregate(value ~ year + fruit, data=df, FUN=function(x) c(sd(x),mean(x))

其中value是一个列，你可以创建一个新的列来放置2021和2022下的值，然后创建一个新的列year，它相应地有2021或2022，在R中，长格式几乎总是要走的路。