max_coverage: Solve the Maximal Covering Location Problem

In njtierney/maxcovr: A Set of Tools For Solving The Maximal Covering Location Problem

Solve the Maximal Covering Location Problem

Description

max_coverage solves the binary optimisation problem known as the "maximal covering location problem" as described by Church (http://www.geo .ucsb.edu/~forest/G294download/MAX_COVER_RLC_CSR.pdf). This package was implemented to make it easier to solve this problem in the context of the research initially presented by Chan et al (http://circ.ahajournals.org/content/127/17/1801.short) to identify ideal locations to place AEDs.

Usage

max_coverage(
  existing_facility,
  proposed_facility,
  user,
  distance_cutoff,
  n_added,
  d_existing_user = NULL,
  d_proposed_user = NULL,
  solver = "glpk"
)

Arguments

existing_facility	data.frame containing the facilities that are already in existing, with columns names lat, and long.
proposed_facility	data.frame containing the facilities that are being proposed, with column names lat, and long.
user	data.frame containing the users of the facilities, along with column names lat, and long.
distance_cutoff	numeric indicating the distance cutoff (in metres) you are interested in. If a number is less than distance_cutoff, it will be 1, if it is greater than it, it will be 0.
n_added	the maximum number of facilities to add.
d_existing_user	Optional distance matrix between existing facilities and users. Default distances are direct (geospherical ellipsoidal) distances; this allows alternative measures such as street-network distances to be submitted (see Examples).
d_proposed_user	Option distance matrix between proposed facilities and users (see Examples).
solver	character "glpk" (default) or "lpSolve". "gurobi" is currently in development, see https://github.com/njtierney/maxcovr/issues/25.

Value

dataframe of results

Examples

library(dplyr)

# already existing locations
york_selected <- york |> filter(grade == "I")

# proposed locations
york_unselected <- york |> filter(grade != "I")

mc_result <- max_coverage(existing_facility = york_selected,
                          proposed_facility = york_unselected,
                          user = york_crime,
                          distance_cutoff = 100,
                          n_added = 20)

mc_result

summary(mc_result)

# get the facilities chosen
mc_result$facility_selected

# get the users affected
mc_result$user_affected

# get the summaries
mc_result$summary

# Example of street-network distance calculations
## Not run: 
library(dodgr)
net <- dodgr_streetnet_sf ("york england") |>
    weight_streetnet (wt_profile = "foot")

from <- match_points_to_graph (v, york_selected[, c ("long", "lat")])
to <- match_points_to_graph (v, york_crime[, c ("long", "lat")])
d_existing_user <- dodgr_dists (net, from = from, to = to)

from <- match_points_to_graph (v, york_unselected[, c ("long", "lat")])
d_proposed_user <- dodgr_dists (net, from = from, to = to)

mc_result <- max_coverage(existing_facility = york_selected,
                          proposed_facility = york_unselected,
                          user = york_crime,
                          distance_cutoff = 100,
                          n_added = 20,
                          d_existing_user = d_existing_user,
                          d_proposed_user = d_proposed_user)


## End(Not run)

njtierney/maxcovr documentation built on Nov. 27, 2024, 7:51 a.m.