温馨提示:本文翻译自stackoverflow.com,查看原文请点击:dataframe - how to subset the rows in a data frame with coordinates based on closest location in R
dataframe r data-manipulation

dataframe - 如何使用基于R中最近位置的坐标对数据框中的行进行子集化

发布于 2020-03-28 23:25:44

我有一个较大的数据框(150000行),其中的X和Y作为df1之类的坐标,如下所示:

df1 <- data.frame(X = c(7.48, 7.82, 8.15, 8.47, 8.80, 9.20, 9.51, 9.83, 10.13, 10.59, 7.59, 8.06, 8.39, 8.87, 9.26, 9.64, 10.09, 10.48, 10.88, 11.45), 
              Y = c(49.16, 48.78, 48.40, 48.03, 47.65, 47.24, 46.87, 46.51, 46.15, 45.73, 48.70, 48.18, 47.72, 47.20, 46.71, 46.23, 45.72, 45.24, 44.77, 44.23), 
              ID = c("B_1", "B_1", "B_1", "B_1", "B_1", "B_1", "B_1", "B_1", "B_1", "B_1", "B_1_2", "B_1_2", "B_1_2", "B_1_2", "B_1_2", "B_1_2", "B_1_2", "B_1_2", "B_1_2", "B_1_2"), 
              TI = c(191.31, 191.35, 191.39, 191.44, 191.48, 191.52, 191.56, 191.60, 191.64, 191.69, 1349.93, 1349.97, 1350.01, 1350.05, 1350.09, 1350.14, 1350.18, 1350.22, 1350.26, 1350.30))

在ID列中,我有大约100-200个唯一ID,在每个唯一ID中,我都有200-300个数据点(行)

我有另一个像df2的数据框,如下所示:

df2 <- data.frame(X = c(7.62,  8.25,  8.95,  9.71,  10.23), 
              Y = c(49.06,  48.30,  47.55,  46.77,  46.25))

现在,基于df2中的每一行,即x1和y1,我想针对唯一的ID找出df1中最接近的XY,如下所示:

 df3 <- 

 X1    Y1   ID1     TI1     X2     Y2    ID2     TI2     X3     Y3    ID3     TI3    X4     Y4    ID4     TI4     X5      Y5       ID5      TI5
7.48 49.16  B_1    191.31  8.15  48.40   B_1    191.39  8.80  47.65   B_1    191.48  9.51  46.87  B_1    191.56   10.13   46.15   B_1     191.64
7.59 48.70  B_1_2  1349.93 8.06  48.18   B_1_2  1349.97 8.87  47.20   B_1_2  1350.05 9.26  46.71  B_1_2  1350.09  10.09   45.72   B_1_2   1350.18

我已经尝试使用以下代码:

dist12 <- function(row){
dists <- (row[["X"]] - df2$X)^2 + (row[["Y"]]- df2$Y)^2
return(cbind(df2[which.min(dists),], distance = min(dists)))
}
df3 <- cbind(df1, do.call(rbind, lapply(1:nrow(df1), function(x) dist12(df1[x,])))) 

该代码通过计算df1和df2数据帧中行之间的距离并组合df1和df2来找到行之间的最小距离。根据此代码,它会将df2的单个XY分配给df1中的多行。但是df2中的单行(XY)只能分配给唯一ID中的行之一。

寻找代码以获取预期的输出(df3),如上所述

提前致谢

查看更多

查看更多

提问者
Kumar
被浏览
16
user2974951 2020-01-31 18:01

这是对data.table性能的一种了解。

    library(data.table)
    df1=as.data.table(df1)
    do.call(cbind,
      apply(df2,1,function(i){
        df1[,d:=(df1$X-i[1])^2+(df1$Y-i[2])^2]
        df1[df1[,.I[d==min(d)],by=ID]$V1]
      })
    )

      X     Y    ID      TI      d    X     Y    ID      TI      d    X     Y    ID      TI      d
1: 7.48 49.16   B_1  191.31 0.0296 8.15 48.40   B_1  191.39 0.0200 8.80 47.65   B_1  191.48 0.0325
2: 7.59 48.70 B_1_2 1349.93 0.1305 8.06 48.18 B_1_2 1349.97 0.0505 8.87 47.20 B_1_2 1350.05 0.1289
      X     Y    ID      TI      d     X     Y    ID      TI      d
1: 9.51 46.87   B_1  191.56 0.0500 10.13 46.15   B_1  191.64 0.0200
2: 9.26 46.71 B_1_2 1350.09 0.2061 10.09 45.72 B_1_2 1350.18 0.3005