Search code examples
rlistdataframegeocodingggmap

R - ggmap - calculate shortest distance between cities via geocoding

I have a list of cities and related information that I've placed in a dataframe, like so:

library(plyr)
library(dplyr)
library(ggmap)
library(Imap)

cities <- c("washington, dc", "wilmington, de", "amarillo, tx", 
            "denver, co", "needham, ma", "philadelphia, pa", 
            "doylestown, pa", "galveston, tx", "tuscaloosa, al", 
            "hollywood, fl"
            )

id <- c(156952, 154222, 785695, 154423, 971453, 149888, 1356987,
        178946, 169944, 136421)

month <- c(201811, 201811, 201912, 201912, 202005, 202005, 
           202005, 202106, 202106, 202106 )

category<- c("home", "work", "home", "home", "home", "work",
             "cell", "home", "work", "cell")

places <- data.frame(cities, id, category, month)

Using the Imap and ggmap packages, I can retrieve the longitude and latitudes for each city:

lat <- geocode(location = places$cities, source = "google")$lat
lon <- geocode(location = places$cities, source = "google")$lon

places <- cbind(places, lat, lon)

What I would like to do is the following:

  1. Calculate the distance between each city by month and category
  2. return the second shortest distance and the corresponding city and id in separate columns in places

I wrote a for loop to calculate the distances:

for (i in 1:nrow(places)) {




dist_list[[i]] <- gdist(lon.1 = places$lon[i], 
                          lat.1 = places$lat[i], 
                          lon.2 = places$lon, 
                          lat.2 = places$lat,
                          units="miles")
  
}

Which produces the following data:

dput(dist_list)
list(c(0, 98.3464717885451, 1386.25425677199, 1489.87718040776, 
383.083760289456, 123.232894969413, 140.284537078237, 1209.23510542932, 
706.670452283757, 906.79542720295), c(98.4762434610638, 0, 1472.06660056474, 
1560.93398322985, 285.23618862797, 24.9195071209828, 44.8853561530985, 
1308.60741637919, 805.755084908157, 983.102810248198), c(1389.07354011351, 
1472.06660056474, 0, 356.573530670257, 1712.29111612461, 1493.39302974566, 
1497.2125164277, 579.329313217289, 827.577713357261, 1434.82691622332
), c(1492.80130415651, 1560.93398322985, 356.573530670257, 0, 
1761.3773163288, 1578.71125031146, 1576.80713231756, 923.725006795209, 
1067.04809350934, 1717.32991551111), c(383.551997010915, 285.23618862797, 
1712.29111612461, 1761.3773163288, 0, 260.382178510916, 243.947043197789, 
1588.85470703957, 1088.38640303169, 1230.47219244291), c(123.395655314093, 
24.9195071209827, 1493.39302974566, 1578.71125031146, 260.382178510916, 
0, 24.7382114555287, 1333.29925285915, 830.581742827321, 1002.94777739349
), c(140.431447025301, 44.8853561530986, 1497.2125164277, 1576.80713231756, 
243.947043197789, 24.7382114555285, 0, 1346.44527983873, 844.827513981938, 
1026.98263808807), c(1211.16392416136, 1308.60741637919, 579.329313217289, 
923.725006795209, 1588.85470703957, 1333.29925285915, 1346.44527983873, 
0, 505.292529136012, 925.512554201542), c(707.73957320737, 805.755084908157, 
827.577713357261, 1067.04809350934, 1088.38640303169, 830.581742827321, 
844.827513981938, 505.292529136012, 0, 666.837848781548), c(906.880841903584, 
983.102810248198, 1434.82691622332, 1717.32991551111, 1230.47219244291, 
1002.94777739349, 1026.98263808807, 925.512554201542, 666.837848781548, 
0))

The desired result would look like this (first row):

cities          id         category  month      lat        lon   min.dist  closest city  closest city id  
washington, dc  156952     home      201811 38.90719  -77.03687   98.34647  wilmington, de  154222

And via the nth function in Rfast I can get the second smallest distance

nth(dist_list[[1]], 2)

The problem I have is I don't know how to connect the info from the list to the df places. Any help or suggestions would be greatly appreciated.

Solution 
  • # get min distance:
min_d <- sapply(dist_list, function(x) sort(x)[2])
places$min_dist <- min_d
# index:
i <- sapply(dist_list, function(x) which(sort(x)[2] == x))
# add name:
places$min_name <- places$cities[i]

    With grouping:

    # prepare dist matrix outside loop
m <- t(as.data.frame(dist_list))
row.names(m) <- NULL
diag(m) <- NA

# create grouping variable:
gv <- as.integer(factor(places$month)) # or:
# gv <- as.integer(factor(paste(places$month, places$category)))

# set distance to NA if not in relevant group:
i <- sapply(gv, function(x) gv == x)
m[!i] <- NA

l <- sapply(as.data.frame(t(m)), function(x) {
  if (all(is.na(x))) return(list(NA, NA))
  mv <- min(x, na.rm = T)
  i <- which(x == mv)
  list(mv, i)
})
l
places <- cbind(places, min_dist = unlist(l[1, ]), min_nr = unlist(l[2, ]))

places$min_name <- places$cities[places$min_nr] # add name
places$min_id <- places$id[places$min_nr] # add id
places

    result:

                  cities      id category  month  min_dist min_nr         min_name  min_id
V1    washington, dc  156952     home 201811  98.34647      2   wilmington, de  154222
V2    wilmington, de  154222     work 201811  98.47624      1   washington, dc  156952
V3      amarillo, tx  785695     home 201912 356.57353      4       denver, co  154423
V4        denver, co  154423     home 201912 356.57353      3     amarillo, tx  785695
V5       needham, ma  971453     home 202005 243.94704      7   doylestown, pa 1356987
V6  philadelphia, pa  149888     work 202005  24.73821      7   doylestown, pa 1356987
V7    doylestown, pa 1356987     cell 202005  24.73821      6 philadelphia, pa  149888
V8     galveston, tx  178946     home 202106 505.29253      9   tuscaloosa, al  169944
V9    tuscaloosa, al  169944     work 202106 505.29253      8    galveston, tx  178946
V10    hollywood, fl  136421     cell 202106 666.83785      9   tuscaloosa, al  169944