空间数据/对在研发邻居计算指标 [英] spatial data / compute metrics on neighbors in R

问题描述

我的形式（XBIN，要用ybin，值）的二维空间数据。 例如：

I have the 2D spatial data in the form (xBin, yBin, value). e.g.:

DT = data.table(x=c(rep(1,3),rep(2,3),rep(3,3)),y=rep(c(1,2,3),3),value=100*c(1:9))

有关每个箱我想计算的变量价值的总和所有邻国箱。 甲仓被认为是邻居如果其两个指数 - x和y是在一个单元内，从当前二进制

For each bin I want to compute the sum of variable "value" over all neighboring bins. A bin is considered a neighbor if both of its indices - x and y are within one unit from the current bin

例如。对于x = 2，Y = 2，我想计算

e.g. for x=2, y=2, I want to compute

valueNeighbors(x=2,y=2) = value(x=1,y=1)+value(1,2)+value(1,3)
+value(2,1)+value(2,3)
+value(3,1)+value(3,2)+value(3,3)

我的真实数据有〜1000 ^ 2箱，我怎么能做到这一点有效？

My real data has ~1,000^2 bins, how can I do this efficiently?

推荐答案

也许与栅格

X <- matrix(1:20, 4)
r <- raster(X)
r
agg <- as.matrix(focal(r,matrix(1,3,3),sum, pad = T, padValue = 0))
agg

     [,1] [,2] [,3] [,4] [,5]
[1,]   14   33   57   81   62
[2,]   24   54   90  126   96
[3,]   30   63   99  135  102
[4,]   22   45   69   93   70

哪种方法更快地为大型数据集？

Which method is the faster for large datasets?

X <- matrix(1:1000000, 1000)
S <- matrix(NA, nrow(X), ncol(X))
r <- raster(X)

system.time(
as.matrix(focal(r,matrix(1,3,3),sum, pad = T, padValue = 0))
)
user  system elapsed 
0.39    0.08    0.47 

通过一个1000×1000矩阵我无法使用提出Winsemius（Win 7的64位8GB RAM）的解决方案reasobable时间内得到的结果。

With a 1000x1000 matrix I was unable to get a result within a reasobable time using the solution proposed by Winsemius (Win 7 x64 8GB RAM)

这篇关于空间数据/对在研发邻居计算指标的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！