变异以错误的顺序返回数据 dplyr 这是一个错误吗？

Question

我遇到了一个问题，其中 dplyr 中的 mutate() 返回结果顺序错误。我对mutate 的调用使用来自现有列的数据作为输入，但返回的结果就像数据在mutate 之前排序一样排列。

我的具体问题使用dataRetrieval 包从网络获取 USGS/NWIS 数据。在此示例中，我根据站点 ID 检索站点名称。在 `dataRetreival 包中，站点 ID 是存储为字符的数字代码。

library(dataRetrieval)
library(dplyr)

Gauges <- tibble( Name = c("Twisp", "Chewuch", "Andrews" ,"Met@Winthrop", "Met@Twisp", "Met@Pateros", "Met@Goat"),
                  ID = c("12448998" , "12448000","12447390", "12448500" ,"12449500","12449950" , "12447383")
)

## This works correctly with each of the station numbers
readNWISsite(Gauges$ID[1])$station_nm
# [1] "TWISP RIVER NEAR TWISP, WA"

## This does not work correctly
## Order is not right! Station does not correspond with ID  !!
Gauges%>%
      mutate(Station = readNWISsite(ID)$station_nm)

# # A tibble: 7 x 3
# Name         ID       Station                                      
# <chr>        <chr>    <chr>                                        
# 1 Twisp        12448998 METHOW RIVER ABOVE GOAT CREEK NEAR MAZAMA, WA
# 2 Chewuch      12448000 ANDREWS CREEK NEAR MAZAMA, WA                
# 3 Andrews      12447390 CHEWUCH RIVER AT WINTHROP, WA                
# 4 Met@Winthrop 12448500 METHOW RIVER AT WINTHROP, WA                 
# 5 Met@Twisp    12449500 TWISP RIVER NEAR TWISP, WA                   
# 6 Met@Pateros  12449950 METHOW RIVER AT TWISP, WA                    
# 7 Met@Goat     12447383 METHOW RIVER NEAR PATEROS, WA    

## This works, returning the correct site associated with the gauge number
Gauges%>%
      arrange(ID) %>%
      mutate(Station = readNWISsite(ID)$station_nm)
# # A tibble: 7 x 3
# Name         ID       Station                                      
# <chr>        <chr>    <chr>                                        
# 1 Met@Goat     12447383 METHOW RIVER ABOVE GOAT CREEK NEAR MAZAMA, WA
# 2 Andrews      12447390 ANDREWS CREEK NEAR MAZAMA, WA                
# 3 Chewuch      12448000 CHEWUCH RIVER AT WINTHROP, WA                
# 4 Met@Winthrop 12448500 METHOW RIVER AT WINTHROP, WA                 
# 5 Twisp        12448998 TWISP RIVER NEAR TWISP, WA                   
# 6 Met@Twisp    12449500 METHOW RIVER AT TWISP, WA                    
# 7 Met@Pateros  12449950 METHOW RIVER NEAR PATEROS, WA  

为什么mutate 在进程中间重新排列数据？或者，这里发生了什么？

Answer 1

要了解发生了什么，不要只提取“station_nm”，还要获取“site_no”

library(dplyr)
library(dataRetrieval)
readNWISsite(Gauges$ID)[c('site_no', 'station_nm')]
#site_no                                    station_nm
#1 12447383 METHOW RIVER ABOVE GOAT CREEK NEAR MAZAMA, WA
#2 12447390                 ANDREWS CREEK NEAR MAZAMA, WA
#3 12448000                 CHEWUCH RIVER AT WINTHROP, WA
#4 12448500                  METHOW RIVER AT WINTHROP, WA
#5 12448998                    TWISP RIVER NEAR TWISP, WA
#6 12449500                     METHOW RIVER AT TWISP, WA
#7 12449950                 METHOW RIVER NEAR PATEROS, WA

这里，“site_no”是根据“ID”的整数值排序的。为了纠正这个问题，我们可以使用 rowwise 一次将函数应用于每个“ID”。

Gauges %>% 
    rowwise() %>% 
    mutate(Station = readNWISsite(ID)$station_nm)

---

或map 来自purrr

library(purrr)
Gauges %>%
    mutate(Station = map_chr(ID, ~ readNWISsite(.x)$station_nm))
# A tibble: 7 x 3
#  Name         ID       Station                                      
#  <chr>        <chr>    <chr>                                        
#1 Twisp        12448998 TWISP RIVER NEAR TWISP, WA                   
#2 Chewuch      12448000 CHEWUCH RIVER AT WINTHROP, WA                
#3 Andrews      12447390 ANDREWS CREEK NEAR MAZAMA, WA                
#4 Met@Winthrop 12448500 METHOW RIVER AT WINTHROP, WA                 
#5 Met@Twisp    12449500 METHOW RIVER AT TWISP, WA                    
#6 Met@Pateros  12449950 METHOW RIVER NEAR PATEROS, WA                
#7 Met@Goat     12447383 METHOW RIVER ABOVE GOAT CREEK NEAR MAZAMA, WA

---

或者我们提取两列并使用“ID”和“site_no”执行match

Gauges %>% 
          mutate(Station = {
           tmp <- readNWISsite(ID)[c('site_no', 'station_nm')]
              tmp$station_nm[match(ID, tmp$site_no)]})
# A tibble: 7 x 3
#  Name         ID       Station                                      
#  <chr>        <chr>    <chr>                                        
#1 Twisp        12448998 TWISP RIVER NEAR TWISP, WA                   
#2 Chewuch      12448000 CHEWUCH RIVER AT WINTHROP, WA                
#3 Andrews      12447390 ANDREWS CREEK NEAR MAZAMA, WA                
#4 Met@Winthrop 12448500 METHOW RIVER AT WINTHROP, WA                 
#5 Met@Twisp    12449500 METHOW RIVER AT TWISP, WA                    
#6 Met@Pateros  12449950 METHOW RIVER NEAR PATEROS, WA                
#7 Met@Goat     12447383 METHOW RIVER ABOVE GOAT CREEK NEAR MAZAMA, WA