将列转换为 r 中的行

Question

我使用代码

形成了以下数据

test <- data.frame(dis = c(10,20,30,40),dur=c(30,40,60,90),method=c("car","car","Bicycle","Bicycle"),to_lon=c(-1.980,-1.5678,-1.324,-1.456),to_lat=c(55.3009,55.3416,55.1123,55.2234),from_lon=c(-1.4565,-1.3424,-1.4566,-1.1111),from_lat=c(76.8888,65.8999,76.9088,25.3344))

 dis dur  method  to_lon  to_lat from_lon from_lat
1  10  30     car -1.9800 55.3009  -1.4565  76.8888
2  20  40     car -1.5678 55.3416  -1.3424  65.8999
3  30  60 Bicycle -1.3240 55.1123  -1.4566  76.9088
4  40  90 Bicycle -1.4560 55.2234  -1.1111  25.3344

我想转换此数据框，使其在一行中包含 to_lat 和 to_lon，在下一行中包含 from_lat 和 from_lon。其余细节不需要更改，可以复制。期望的结果应该如下

    dis dur method  longitude   latitude
from    10  30  car -1.98   55.3009
to  10  30  car -1.4565 76.8888
from    20  40  car -1.5678 55.3416
to  20  40  car -1.3424 65.8999
from    30  60  Bicycle -1.324  55.1123
to  30  60  Bicycle -1.4566 76.9088
from    40  90  Bicycle -1.456  55.2234
to  40  90  Bicycle -1.1111 25.3344

任何帮助将不胜感激。

谢谢。

Answer 1

我们可以使用 data.table 中的 melt，它可以包含多个 measure 列。

library(data.table)
dM <- melt(setDT(test), measure=patterns('lon', 'lat'), 
          value.name=c('longitude', 'latitude'))
#change the 'variable' column from numeric index to 'from/to'
dM[, variable:= c('from', 'to')[variable]]
#create a sequence column grouped by 'variable'
dM[,i1:= 1:.N ,variable]
#order based on the 'i1'
res <- dM[order(i1)][,i1:=NULL]
res
#    dis dur  method variable longitude latitude
#1:  10  30     car     from   -1.9800  55.3009
#2:  10  30     car       to   -1.4565  76.8888
#3:  20  40     car     from   -1.5678  55.3416
#4:  20  40     car       to   -1.3424  65.8999
#5:  30  60 Bicycle     from   -1.3240  55.1123
#6:  30  60 Bicycle       to   -1.4566  76.9088
#7:  40  90 Bicycle     from   -1.4560  55.2234
#8:  40  90 Bicycle       to   -1.1111  25.3344

Answer 2

这可能不是最优雅的解决方案，但它应该可以工作并且可以理解：

我们将数据分成两个数据帧：一个包含 'from' 经度和纬度数据（称为 testF），另一个包含 'to' 数据（称为 test）。然后我们使用 rbind 将 'testF' 的行插入 'test'.

中的适当位置

test <- data.frame(dis = c(10,20,30,40),dur=c(30,40,60,90),method=c("car","car","Bicycle","Bicycle"),to_lon=c(-1.980,-1.5678,-1.324,-1.456),to_lat=c(55.3009,55.3416,55.1123,55.2234),from_lon=c(-1.4565,-1.3424,-1.4566,-1.1111),from_lat=c(76.8888,65.8999,76.9088,25.3344))

testF <- test[,c(1:3,6,7)]
names(testF)[4:5] <- c("lonitude", "latitude")
test <- test[,1:5]
names(test)[4:5] <- c("lonitude", "latitude")

for(i in dim(test)[1]:1) {
  test <- rbind(test[1:i,], testF[i,], test[-(1:i),])
}

Answer 3

这是使用包 tidyr（一种流行的数据处理包）的替代方法，它避免了 for 循环。

library(tidyr)

test <- data.frame(dis = c(10,20,30,40),dur=c(30,40,60,90),method=c("car","car","Bicycle","Bicycle"),to_lon=c(-1.980,-1.5678,-1.324,-1.456),to_lat=c(55.3009,55.3416,55.1123,55.2234),from_lon=c(-1.4565,-1.3424,-1.4566,-1.1111),from_lat=c(76.8888,65.8999,76.9088,25.3344))
test$id <- 1:dim(test)[1]

# gather latitude columns
d1 <- gather(data = test, 
             key = direction, 
             value = latitude, 
             to_lat, from_lat)

# gather longitude columns
d2 <- gather(data = test, 
             key = direction, 
             value = longitude, 
             to_lon, from_lon)

d3 <- cbind(d1[,c("direction","dis","dur","method","latitude")],d2[,c("longitude","id"),drop=FALSE])

# Create names
dir <- unlist(strsplit(d3$direction,"_"))
dir <- dir[seq(from = 1, to = length(dir), by = 2)]

# Factor and sort
d3$direction <- factor(dir)
d3[order(d3$id),]

将列转换为 r 中的行

Converting columns into rows in r

r

transform

reshape2

tidyr