如何在 R 中按条件从数据框中子集相同的行？答案

【问题标题】：How to subset identical rows from dataframe by condition in R?如何在 R 中按条件从数据框中子集相同的行？
【发布时间】：2022-08-21 18:51:03
【问题描述】：

从这个数据框中，我想对具有相同 \"model\"、\"Country\" 和 \"year\" 值但不同 \"factor\" 的行进行子集化

model <- c(\"A\",\"B\",\"C\",\"A\",\"A\",\"C\",\"B\",\"A\")
country <- c(\"Italy\",\"Germany\",\"Norway\",\"Italy\",\"France\",\"Germany\",\"Norway\",\"Italy\")
year <- c(2016,2016,2016,2016,2015,2015,2015,2015)
value <-  c(14,24,11,34,73,11,33,22)
factor <- c(\"NEW\",\"OLD\",\"OLD\",\"OLD\",\"OLD\",\"OLD\",\"NEW\",\"NEW\")

df <- data.frame(model,country,year,value, factor)

  model country year value factor
1     A   Italy 2016    14    NEW
2     B Germany 2016    24    OLD
3     C  Norway 2016    11    OLD
4     A   Italy 2016    34    OLD
5     A  France 2015    73    OLD
6     C Germany 2015    11    OLD
7     B  Norway 2015    33    NEW
8     A   Italy 2015    22    NEW

预期输出：

1     A   Italy 2016    22    NEW
4     A   Italy 2016    34    OLD

第 1 行的输出不应该是 14 而不是 22？

标签： dataframe loops for-loop dplyr tidyverse

【解决方案1】：

你可以做的是创建一个列，根据组（型号、国家和年份）告诉重复的数量，然后 filter 如下所示：

model <- c("A","B","C","A","A","C","B","A")
country <- c("Italy","Germany","Norway","Italy","France","Germany","Norway","Italy")
year <- c(2016,2016,2016,2016,2015,2015,2015,2015)
value <-  c(14,24,11,34,73,11,33,22)
factor <- c("NEW","OLD","OLD","OLD","OLD","OLD","NEW","NEW")

df <- data.frame(model,country,year,value, factor)

library(dplyr)

df %>% 
  group_by(model, country, year) %>% 
  mutate(number_dups = n()) %>% 
  filter(number_dups > 1) %>%
  select(-number_dups) %>%
  ungroup()
#> # A tibble: 2 × 5
#>   model country  year value factor
#>   <chr> <chr>   <dbl> <dbl> <chr> 
#> 1 A     Italy    2016    14 NEW   
#> 2 A     Italy    2016    34 OLD

^{由reprex package (v2.0.1) 于 2022 年 8 月 12 日创建}

请注意：我假设预期的输出应该是 14 而不是 22。

【讨论】：

【解决方案2】：

在基础 R 中，您可以使用 subset + ave

> subset(df, ave(seq_along(year), model, country, year, FUN = length) > 1)
  model country year value factor
1     A   Italy 2016    14    NEW
4     A   Italy 2016    34    OLD

或者，试试Filter + split

> Filter(function(x) nrow(x) > 1, split(df, ~ model + country + year))
$A.Italy.2016
  model country year value factor
1     A   Italy 2016    14    NEW
4     A   Italy 2016    34    OLD

或者

> do.call(rbind, unname(Filter(function(x) nrow(x) > 1, split(df, ~ model + country + year))))
  model country year value factor
1     A   Italy 2016    14    NEW
4     A   Italy 2016    34    OLD

【讨论】：

【解决方案3】：

您可以按model、country 和year 进行分组，然后使用filter 保留具有model、country 和year 重复值的行。然后，为了确保有不同的因子值，我们使用n_distinct 来确保factor 中至少有2 个不同的值（即，该组必须有NEW 和OLD）。

library(dplyr)
    
df %>%
  group_by(model, country, year) %>%
  filter(n() > 1 & n_distinct(factor) > 1)

输出

  model country  year value factor
  <chr> <chr>   <dbl> <dbl> <chr> 
1 A     Italy    2016    14 NEW   
2 A     Italy    2016    34 OLD

附加示例：每组中的多个因素

如果您只想在每个组中保留 1 行不同的因子（即 NEW 和 OLD），那么我们可以添加 factor 作为附加分组变量并使用 slice 仅保留第一行每组。

df2 <- structure(list(model = c("A", "B", "C", "A", "A", "C", "B", "A"
), country = c("Italy", "Germany", "Norway", "Italy", "France", 
"Germany", "Norway", "Italy"), year = c(2016, 2016, 2016, 2016, 
2015, 2015, 2015, 2016), value = c(14, 34, 11, 34, 33, 11, 33, 
22), factor = c("NEW", "OLD", "OLD", "OLD", "OLD", "OLD", "NEW", 
"NEW")), class = "data.frame", row.names = c(NA, -8L))

#  model country year value factor
#1     A   Italy 2016    14    NEW
#2     B Germany 2016    34    OLD
#3     C  Norway 2016    11    OLD
#4     A   Italy 2016    34    OLD
#5     A  France 2015    33    OLD
#6     C Germany 2015    11    OLD
#7     B  Norway 2015    33    NEW
#8     A   Italy 2016    22    NEW

df2 %>%
  group_by(model, country, year) %>%
  filter(n() > 1 & n_distinct(factor) > 1) %>%
  group_by(factor, .add = T) %>%
  slice(1)

输出

  model country  year value factor
  <chr> <chr>   <dbl> <dbl> <chr> 
1 A     Italy    2016    14 NEW   
2 A     Italy    2016    34 OLD

【讨论】：