根据一列中的条件在数据框中创建新变量，从其他列拉出？ （dplyr）

Question

我有以下数据帧：

    df <- structure(list(country = c("Ghana", "Eritrea", "Ethiopia", "Ethiopia", 
"Congo - Kinshasa", "Ethiopia", "Ethiopia", "Ghana", "Botswana", 
"Nigeria"), CommodRank = c(1L, 2L, 3L, 1L, 3L, 1L, 1L, 1L, 1L, 
1L), topCommodInCountry = c(TRUE, FALSE, FALSE, TRUE, FALSE, 
TRUE, TRUE, TRUE, TRUE, TRUE), Main_Commod = c("Gold", "Copper", 
"Nickel", "Gold", "Gold", "Gold", "Gold", "Gold", "Diamonds", 
"Iron Ore")), row.names = c(NA, -10L), class = c("grouped_df", 
"tbl_df", "tbl", "data.frame"), vars = "country", drop = TRUE, indices = list(
    8L, 4L, 1L, c(2L, 3L, 5L, 6L), c(0L, 7L), 9L), group_sizes = c(1L, 
1L, 1L, 4L, 2L, 1L), biggest_group_size = 4L, labels = structure(list(
    country = c("Botswana", "Congo - Kinshasa", "Eritrea", "Ethiopia", 
    "Ghana", "Nigeria")), row.names = c(NA, -6L), class = "data.frame", vars = "country", drop = TRUE, .Names = "country"), .Names = c("country", 
"CommodRank", "topCommodInCountry", "Main_Commod"))

df

            country CommodRank topCommodInCountry Main_Commod
1             Ghana          1               TRUE        Gold
2           Eritrea          2              FALSE      Copper
3          Ethiopia          3              FALSE      Nickel
4          Ethiopia          1               TRUE        Gold
5  Congo - Kinshasa          3              FALSE        Gold
6          Ethiopia          1               TRUE        Gold
7          Ethiopia          1               TRUE        Gold
8             Ghana          1               TRUE        Gold
9          Botswana          1               TRUE    Diamonds
10          Nigeria          1               TRUE    Iron Ore  

我正在尝试添加另一列显示此数据集中每个国家/地区的顶级商品（顶级CommodRank），但我不确定如何。 我可以用'Main_Commod'标记'topcommod'，其中CommodRank == 1，但我想将这个相同的值复制到CommodRank！= 1的情况下。看下面，第3行和第4行的埃塞俄比亚值都应该是'金'。

df %>% mutate(topcommod = ifelse(CommodRank == 1, Main_Commod, 'unknown'))


            country CommodRank topCommodInCountry Main_Commod topcommod
1             Ghana          1               TRUE        Gold      Gold
2           Eritrea          2              FALSE      Copper   unknown
3          Ethiopia          3              FALSE      Nickel   unknown
4          Ethiopia          1               TRUE        Gold      Gold
5  Congo - Kinshasa          3              FALSE        Gold   unknown
6          Ethiopia          1               TRUE        Gold      Gold
7          Ethiopia          1               TRUE        Gold      Gold
8             Ghana          1               TRUE        Gold      Gold
9          Botswana          1               TRUE    Diamonds  Diamonds
10          Nigeria          1               TRUE    Iron Ore  Iron Ore

我理想地寻找一个dplyr解决方案，我可以添加到现有的长系列管道％>％函数调用，但任何解决方案都会有所帮助。

Answer 1

IIUC，有多种方法可以做到这一点，例如：

df %>% mutate(topCom = if(!any(topCommodInCountry)) "unknown" 
                       else Main_Commod[which.max(topCommodInCountry)])

# A tibble: 10 x 5
# Groups:   country [6]
   country          CommodRank topCommodInCountry Main_Commod topCom  
   <chr>                 <int> <lgl>              <chr>       <chr>   
 1 Ghana                     1 TRUE               Gold        Gold    
 2 Eritrea                   2 FALSE              Copper      unknown 
 3 Ethiopia                  3 FALSE              Nickel      Gold    
 4 Ethiopia                  1 TRUE               Gold        Gold    
 5 Congo - Kinshasa          3 FALSE              Gold        unknown 
 6 Ethiopia                  1 TRUE               Gold        Gold    
 7 Ethiopia                  1 TRUE               Gold        Gold    
 8 Ghana                     1 TRUE               Gold        Gold    
 9 Botswana                  1 TRUE               Diamonds    Diamonds
10 Nigeria                   1 TRUE               Iron Ore    Iron Ore

关于OP在评论中如何处理多个顶级商品关系的问题，您可以执行以下操作：

df %>% 
  mutate(topCom = if(!any(topCommodInCountry)) "unknown" 
              else paste(unique(Main_Commod[topCommodInCountry]), collapse = "/"))

如果一个国家/地区有多个唯一的顶级商品，它们将粘贴到一个字符串中，用/分隔。

Answer 2

dplyr另一种模式......

df %>% arrange(CommodRank) %>%
    mutate(topCommod = Main_Commod[1])

Answer 3

这不是一个答案，但是从@docendo discimus回答中学到了很多东西，我花了一秒钟来理解“if negative”（ !any(topCommodInCountry) ），我想知道它是否只有我还是会让我的计算机更多这样做:)

使用相同的数据集，我检查了使if else积极的想法。 首先，我测试了identical的两种解决方案之间：

identical(
  #Negative
  df %>% 
    mutate(topCom = if(!any(topCommodInCountry)) "unknown" 
           else Main_Commod[which.max(topCommodInCountry)]), 
  #Positive
  df %>% 
    mutate(topCom = if(any(topCommodInCountry)) Main_Commod[which.max(topCommodInCountry)] 
           else "unknown"))

[1] TRUE

接下来，我测试了两者的基准：

require(rbenchmark)

benchmark("Negative" = {
  df %>% 
    mutate(topCom = if(!any(topCommodInCountry)) "unknown" 
           else Main_Commod[which.max(topCommodInCountry)])
},
"Positive" = {
  df %>% 
    mutate(topCom = if(any(topCommodInCountry)) Main_Commod[which.max(topCommodInCountry)] 
           else  "unknown")
},
replications = 10000,
columns = c("test", "replications", "elapsed",
            "relative", "user.self", "sys.self"))

差异并不大，但我假设随着更大的数据集，它会增加。

      test replications elapsed relative user.self sys.self
1 Negative        10000   12.59    1.015     12.44        0
2 Positive        10000   12.41    1.000     12.30        0