How to calculate geographical distance between multiple individuals at same timestep in R

Question

I have a matrix (22467 rows and 4 columns) of x and y GPS locations (decimal degrees) for multiple time steps (one hour) for multiple individuals (ID, n=13). Example of dataset (saved as csv file):

ID  Time  x  y  
98427  01:00  43.97426  -59.56677

98427  02:00  43.97424  -60.56970

98428  01:00  43.97434  -60.52222

98428  02:00  43.97435  -59.24356

98429  01:00  43.97657  -59.36576

98429  02:00  43.97432  -59.98674

I would like to calculate the distance between each individual, for all combinations, at each time step. Thus, at Time = 01:00, distance between 98427 and 98428, 98427 and 98429, 98428 and 98429, etc. How can I do this in R?

Answer 1

library(plyr)

data = iris
data = data[c(1:5, 81:85, 141:145), 3:5]
data$time = rep(1:5, 3)

dlply(data, .(time), function(x) {dist(x[ , 1:2])})

I just played with iris dataset, but methodology is very simillar.
1. Split the data by time
2. Use column x and y and pass to dist() function which returns matrix of distances
3. Store each as list

Then you can pull values from list, which has each entry named as time.

---

Update : Sorry about naively thinking as euclidean distance. Here's somewhat crude implementation of Haversine distance.

library(geosphere)

havdist = function(x) {
  n = dim(x)[1]
  res = matrix(NA, nrow = n, ncol = n)
  for (i in 1:n) {
    k = 1
    for (j in k:n) {
      res[i, j] = res[j, i] = distHaversine(a[i, ], a[j, ])
    }
    n = n - 1
    k = k + 1
   }
  return(res)
}

Then supply havdist instead of dist in above dlply method.