include ability to add dist matrices for #77

[mpadge]

Description

Adds 2 more parameters to the max_coverage function:

1. d_existing_user; and
2. d_proposed_user

These are explained in updated man entries

Example

Here's a reprex:

setwd ("/data/mega/code/forks/maxcovr")
devtools::load_all (".", export_all = TRUE)
#> Loading maxcovr
devtools::load_all ("../../repos/atfutures/dodgr", export_all = FALSE)
#> Loading dodgr
suppressMessages (library (dplyr))
dodgr_cache_off ()
net <- dodgr_streetnet_sc ("york, england") %>%
    weight_streetnet (wt_profile = "foot")
#> Loading required namespace: geodist
v <- dodgr_vertices (net)

existing_facility <- york_selected <- york %>% filter(grade == "I")
proposed_facility <- york_unselected <- york %>% filter(grade != "I")
user <- york_crime 

from <- match_points_to_graph (v, existing_facility [, c ("long", "lat")])
to <- match_points_to_graph (v, user [, c ("long", "lat")])
d_existing <- dodgr_dists (net, from = from, to = to)

from <- match_points_to_graph (v, proposed_facility [, c ("long", "lat")])
d_proposed <- 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)
mc_result_dmats <- max_coverage(existing_facility = york_selected,
                                proposed_facility = york_unselected,
                                user = york_crime,
                                distance_cutoff = 100,
                                n_added = 20,
                                d_existing_user = d_existing,
                                d_proposed_user = d_proposed)
summary (mc_result)
#> 
#> ------------------------------------------- 
#> Model Fit: maxcovr fixed location model 
#> ------------------------------------------- 
#> Distance Cutoff: 100m 
#> Facilities: 
#>     Added:       20 
#> Coverage (Previous): 
#>     # Users:     540    (339) 
#>     Proportion:  0.2977 (0.1869) 
#> Distance (m) to Facility (Previous): 
#>     Avg:         886 (1400) 
#>     SD:          986 (1597) 
#> -------------------------------------------
summary (mc_result_dmats)
#> 
#> ------------------------------------------- 
#> Model Fit: maxcovr fixed location model 
#> ------------------------------------------- 
#> Distance Cutoff: 100m 
#> Facilities: 
#>     Added:       20 
#> Coverage (Previous): 
#>     # Users:     490    (339) 
#>     Proportion:  0.2701 (0.1869) 
#> Distance (m) to Facility (Previous): 
#>     Avg:         1092 (1400) 
#>     SD:          1193 (1597) 
#> -------------------------------------------

^{Created on 2019-11-08 by the reprex package (v0.3.0)}

All measures are reduced, simply because 100m of street distance is equivalent to using a lower value of dist_threshold in the first model. Alternatively, just put in dist_threshold = 125 in the model with the distance matrices and you'll get the same number of users covered (540). Distances to facilities nevertheless remain greater, again coz of street distances.

Tests

... sorry, ain't got no tests, because that requires some kind of distance matrices to be used. We can discuss that later.

[njtierney]

🎉 Looks good to me, thanks Mark! :)

[mpadge]

Pleasure. I'm going to update dodgr so the facility data can just be submitted straight to dodgr_dists, without needing the separate match_points_to_graph calls. I'll get another dodgr on CRAN pretty soon, and then be able to simplify the examples