如果多个列匹配，则从一个数据帧复制到另一个

Question

我有两个信息相似的不同数据帧。一个(df2(有一个更好的UNIQFIREID列表，第二个(df1(是我需要使用的数据帧，因为它包含我正在处理的形状文件。如果df1的UNIQFIREID是NA，并且两个数据帧之间的多列匹配，在这种情况下是FIRENAME、DISCOVERDATETIME和TOTALACRES，我希望能够将UNIQFIREIDs从df2复制并粘贴到df1中。然后忽略那些没有NA或不匹配的。我在下面放了一些小样本数据帧来处理。

到目前为止，我所尝试的方法，比如使用merge、match、join和ifelse方法，只是造成了一堆复杂的混乱，因为我不确定自己在做什么。我在Stack Overflow上发现了一些类似的问题，但它们要简单得多，我找不到组合方法的方法。如有任何建议，我们将不胜感激。

df1 <- data.frame(FIRENAME = c("Gold", "Tree", "Tank", "Green_1"),
UNIQFIREID = c("1985-AZASF-000285", NA, "1985-AZASF-000287", "1985-AZASF-000288"),
DISCOVERYDATETIME = c("1985-03-28", "1985-03-29", "1985-03-30", "1985-03-31"),
TOTALACRES = c(60, 70, 80, 90))
df1$DISCOVERYDATETIME <- as.POSIXct(df1$DISCOVERYDATETIME)
df2 <- data.frame(FIRENAME = c("Gold", "Tree", "Tank", "Green_1"),
UNIQFIREID = c("1985-AZASF-000285", "1985-AZASF-000286", "1985-AZASF-000287", "1985-AZASF-000288"),
DISCOVERYDATETIME = c("1985-03-28", "1985-03-29", "1985-03-30", "1985-03-31"),
TOTALACRES = c(60, 70, 80, 90))
df2$DISCOVERYDATETIME <- as.POSIXct(df2$DISCOVERYDATETIME)

这是一堆垃圾，我一直在努力让它发挥作用。我不建议运行任何一个，但这更像是一个例子，看看我做得有多糟糕。

SW_Fire_Perimeters_1985test$UNIQFIREID[is.na(SW_Fire_Perimeters_1985test$UNIQFIREID)] <-
SW_Fire_Occurrences_1985[match(paste(SW_Fire_Perimeters_1985test$DISCOVERYDATETIME, 
SW_Fire_Perimeters_1985test$FIRENAME, 
SW_Fire_Perimeters_1985test$TOTALACRES), 
paste(SW_Fire_Occurrences_1985$DISCOVERYDATETIME, 
SW_Fire_Occurrences_1985$FIRENAME, 
SW_Fire_Occurrences_1985$TOTALACRES)),"UNIQFIREID"]
ifelse(is.na(SW_Fire_Perimeters_1985test$UNIQFIREID), 
SW_Fire_Occurrences_1985[match(paste(SW_Fire_Perimeters_1985test$DISCOVERYDATETIME, 
SW_Fire_Perimeters_1985test$FIRENAME, 
SW_Fire_Perimeters_1985test$TOTALACRES), 
paste(SW_Fire_Occurrences_1985$DISCOVERYDATETIME, SW_Fire_Occurrences_1985$FIRENAME, 
SW_Fire_Occurrences_1985$TOTALACRES)),"UNIQFIREID"])

SW_Fire_Perimeters_1985test$UNIQFIREID2 <- 
SW_Fire_Occurrences_1985[match(paste(SW_Fire_Perimeters_1985test$DISCOVERYDATETIME, 
SW_Fire_Perimeters_1985test$FIRENAME, 
SW_Fire_Perimeters_1985test$TOTALACRES), 
paste(SW_Fire_Occurrences_1985$DISCOVERYDATETIME, SW_Fire_Occurrences_1985$FIRENAME, 
SW_Fire_Occurrences_1985$TOTALACRES)),"UNIQFIREID"]
# Merges two dataframes into fire perimeters dataframe based on "DISCOVERYDATETIME", "FIRENAME", "TOTALACRES" 
# https://docs.tibco.com/pub/enterprise-runtime-for-R/4.0.0/doc/html/Language_Reference/base/merge.html

SW_Fire_Merge_1985 <- merge(SW_Fire_Perimeters_1985, SW_Fire_Occurrences_1985, on = c( "DISCOVERYDATETIME", "FIRENAME", "TOTALACRES"), nomatch = 0L) 

SW_Fire_join_1985 <- full_join(SW_Fire_Perimeters_1985,SW_Fire_Occurrences_1985,
copy = TRUE, 
# by.x = c("DISCOVERYDATETIME", "FIRENAME", "TOTALACRES"),
# by.y = c("DISCOVERYDATETIME", "FIRENAME", "TOTALACRES"),
# all.x = TRUE),
# by.y = c("UNIQFIREID"))
if(is.na(SW_Fire_Merge_1985$UNIQFIREID.x, paste(SW_Fire_Merge_1985$UNIQFIREID.y)))

如果你想看到完整的数据集(14Mb压缩(和我所在的位置，你可以使用以下代码。只需替换"；目录"以及您想要下载数据和打开文件的位置。它选择了1985年的一个较小的集合来与合作

# Insert path to Geospatial data needed, and desired download location
FireH <- download.file("http://www.fs.fed.us/r3/gis/gisdata/Fire_History.zip",  "Directory.../Fire_History.zip")
# Insert File path of downloaded zip file, overwrite is currently enabled for coding purposes,  for exdir insert desired file path for geodatabase.
FireH2 <- unzip("Directory.../Fire_History.zip", overwrite = TRUE, exdir = "Directory...")
# Assigning Geodatabase a name
FireHGDB <- "Direcrory.../Fire_History.gdb"
# Brings Fire perimeters and occurrences out of GDB 
SW_Fire_Perimeters <- st_read(FireHGDB, "FirePerimeter") #require_geomType="wkbPolygon")
SW_Fire_Occurrences <- st_read(FireHGDB, "FireOccurrence") #require_geomType="wkbPolygon")
# Removes invalid naming characters
# https://www.journaldev.com/43690/sub-and-gsub-function-r#the-gsub-function-in-r
SW_Fire_Perimeters$FIRENAME <- gsub(" ", "_", SW_Fire_Perimeters$FIRENAME) 
SW_Fire_Occurrences$FIRENAME <- gsub(" ", "_", SW_Fire_Occurrences$FIRENAME) 
SW_Fire_Perimeters$FIRENAME <- gsub("#", "_", SW_Fire_Perimeters$FIRENAME)
SW_Fire_Occurrences$FIRENAME <- gsub("#", "_", SW_Fire_Occurrences$FIRENAME)
SW_Fire_Perimeters$FIRENAME <- gsub("\.", "", SW_Fire_Perimeters$FIRENAME)
SW_Fire_Occurrences$FIRENAME <- gsub("\.", "", SW_Fire_Occurrences$FIRENAME)
# Removes NAs from fire occurrences UNIQFIREID column
SW_Fire_Occurrences <- SW_Fire_Occurrences[!is.na(SW_Fire_Occurrences$UNIQFIREID),]
# Removes incomplete UNIQFIREIDs for fire occurrences
SW_Fire_Occurrences <- subset(SW_Fire_Occurrences, nchar(as.character(UNIQFIREID)) == 17)
# Removes geometries from fire occurrences so they can be merged to perimeters (Error with two sf objects when merged)
SW_Fire_Occurrences <- st_drop_geometry(SW_Fire_Occurrences)
# Filters tables to only contain FIREYEARs 1985 - 2019
SW_Fire_Perimeters_1985_2019 <- filter(SW_Fire_Perimeters, FIREYEAR >= 1985, FIREYEAR <= 2019)
SW_Fire_Occurrences_1985_2019 <- filter(SW_Fire_Occurrences, FIREYEAR >= 1985, FIREYEAR <= 2019)
# Make a new row (UniqLength) with the string length of UNIQFIREID (it should be 17 characters long)
SW_Fire_Perimeters_1985_2019$UniqLength <- str_count(SW_Fire_Perimeters_1985_2019$UNIQFIREID)
# Set NAs is UniqLength to 0
# https://stackoverflow.com/questions/7279089/replace-all-na-with-false-in-selected-columns-in-r
SW_Fire_Perimeters_1985_2019[c("UniqLength")][is.na(SW_Fire_Perimeters_1985_2019[c("UniqLength")])] <- FALSE
# Replace any UNIQFIREIDs with NA when UNIQFIREID (UniqLength) not equal to 17
# https://stackoverflow.com/questions/56681308/converting-values-to-na-with-conditions-in-r
SW_Fire_Perimeters_1985_2019[SW_Fire_Perimeters_1985_2019$UniqLength !=17,c("UNIQFIREID")] <- NA
# Filter to FIREYEAR 1985 only
SW_Fire_Perimeters_1985 <- filter(SW_Fire_Perimeters_1985_2019, FIREYEAR == 1985)
SW_Fire_Occurrences_1985 <- filter(SW_Fire_Occurrences_1985_2019, FIREYEAR == 1985)

Answer 1

如果我理解正确，你可以。。。

1. 执行完全联接，其中by=是除"UNIQFIREID"之外的所有列

• 结果将保持值从。。。
  • <RESULT>$UNIQFIREID.x中的df1$UNIQFIREID
  • <RESULT>$UNIQFIREID.y中的df2$UNIQFIREID

2. 使用ifelse()(或其亲属(根据需要从<RESULT>$UNIQFIREID.x和<RESULT>$UNIQFIREID.y中提取值，创建一个新的"UNIQFIREID"列
3. 删除CCD_ 11CCD_

您的数据：

df1 <- data.frame(FIRENAME = c("Gold", "Tree", "Tank", "Green_1"),
UNIQFIREID = c("1985-AZASF-000285", NA, "1985-AZASF-000287", "1985-AZASF-000288"),
DISCOVERYDATETIME = c("1985-03-28", "1985-03-29", "1985-03-30", "1985-03-31"),
TOTALACRES = c(60, 70, 80, 90))
df1$DISCOVERYDATETIME <- as.POSIXct(df1$DISCOVERYDATETIME)
df2 <- data.frame(FIRENAME = c("Gold", "Tree", "Tank", "Green_1"),
UNIQFIREID = c("1985-AZASF-000285", "1985-AZASF-000286", "1985-AZASF-000287", "1985-AZASF-000288"),
DISCOVERYDATETIME = c("1985-03-28", "1985-03-29", "1985-03-30", "1985-03-31"),
TOTALACRES = c(60, 70, 80, 90))
df2$DISCOVERYDATETIME <- as.POSIXct(df2$DISCOVERYDATETIME)

使用{base}:

combo_base <- merge(df1, df2, all = TRUE,
by = c("FIRENAME", "DISCOVERYDATETIME", "TOTALACRES"))
combo_base$UNIQFIREID <- ifelse(is.na(combo_base$UNIQFIREID.x), 
combo_base$UNIQFIREID.y, combo_base$UNIQFIREID.x)
combo_base <- combo_base[!is.na(combo_base$UNIQFIREID), 
!names(combo_base) %in% c("UNIQFIREID.x", "UNIQFIREID.y"), 
drop = FALSE]
combo_base
#>   FIRENAME DISCOVERYDATETIME TOTALACRES        UNIQFIREID
#> 1     Gold        1985-03-28         60 1985-AZASF-000285
#> 2  Green_1        1985-03-31         90 1985-AZASF-000288
#> 3     Tank        1985-03-30         80 1985-AZASF-000287
#> 4     Tree        1985-03-29         70 1985-AZASF-000286

使用{data.table}:

library(data.table)
combo_datatable <- merge(
as.data.table(df1), df2, 
by = c("FIRENAME", "DISCOVERYDATETIME", "TOTALACRES"),
all = TRUE
)[, UNIQFIREID := fifelse(is.na(UNIQFIREID.x), UNIQFIREID.y, UNIQFIREID.x)
][!is.na(UNIQFIREID), !c("UNIQFIREID.x", "UNIQFIREID.y")
]
combo_datatable
#>    FIRENAME DISCOVERYDATETIME TOTALACRES        UNIQFIREID
#> 1:     Gold        1985-03-28         60 1985-AZASF-000285
#> 2:  Green_1        1985-03-31         90 1985-AZASF-000288
#> 3:     Tank        1985-03-30         80 1985-AZASF-000287
#> 4:     Tree        1985-03-29         70 1985-AZASF-000286

使用{dplyr}:

library(dplyr, warn.conflicts = FALSE)
combo_dplyr <- df1 %>% 
full_join(df2, by = c("FIRENAME", "DISCOVERYDATETIME", "TOTALACRES")) %>% 
mutate(UNIQFIREID = if_else(is.na(UNIQFIREID.x), UNIQFIREID.y, UNIQFIREID.x)) %>% 
select(-UNIQFIREID.x, -UNIQFIREID.y) %>% 
filter(!is.na(UNIQFIREID))
combo_dplyr
#>   FIRENAME DISCOVERYDATETIME TOTALACRES        UNIQFIREID
#> 1     Gold        1985-03-28         60 1985-AZASF-000285
#> 2     Tree        1985-03-29         70 1985-AZASF-000286
#> 3     Tank        1985-03-30         80 1985-AZASF-000287
#> 4  Green_1        1985-03-31         90 1985-AZASF-000288

卫生检查：

identical(combo_base, as.data.frame(combo_datatable))
#> [1] TRUE
identical(combo_base, combo_dplyr %>% arrange(FIRENAME))
#> [1] TRUE