Geographical distance by group - Applying a function on each pair of rows

Question

I want to calculate the average geographical distance between a number of houses per province.

Suppose I have the following data.

df1 <- data.fram         


        

Answer 1

Solution:

lapply(split(df1, df1$province), function(df){
  df <- Expand.Grid(df[, c("lat", "lon")], df[, c("lat", "lon")])
  mean(distHaversine(df[, 1:2], df[, 3:4]))
})

where Expand.Grid() is taken from https://stackoverflow.com/a/30085602/3502164.

Explanation:

1. Performance

I would avoid using distm() as it transforms a vectorised function distHaversine() into an unvectorised distm(). If you look at the source code you see:

function (x, y, fun = distHaversine) 
{
   [...]
   for (i in 1:n) {
        dm[i, ] = fun(x[i, ], y)
    }
    return(dm)
}

While distHaversine() sends the "whole object" to C, distm() sends the data "row-wise" to distHaversine() and therefore forces distHaversine() to do the same when executing the code in C. Therefore, distm() should not be used. In terms of performance i see more harm using the wrapper function distm() as i see benefits.

2. Explaining the code in "solution":

a) Splitting in groups:

You want to analyse the data per group: province. Splitting into groups can be done by: split(df1, df1$province).

b) Grouping "clumps of columns"

You want to find all unique combinations of lat/lon. First guess might be expand.grid(), but that does not work for mulitple columns. Luckily Mr. Flick took care of this expand.grid function for data.frames in R.

Then you have a data.frame() of all possible combinations and just have to use mean(distHaversine(...)).