Calculate expected distance

[mpadge]

This is described at the start of Appendix D in Saerens et al. (2009) - implement it!

@karpfen one for you: This actually requires an additional modification of the current osmlines_as_network function because edge weights are distances times the weighting system for the preferred type of routing. The routing has to be calculated from these weights, but we also need a final (probabilistic) distance which has to be calculated from the raw distances. So the result of osmlines_as_network needs both of these columns.

The rcpp_router function thus needs to be modified accordingly. Maybe I'll do that, and you can fix up osmlines_as_network?

[karpfen]

That does make sense. I'll get to it tomorrow!

[karpfen]

Allright, I finished adapting osmlines_as_network. You can pass a road type as a parameter and the distance is weighted according to the Routino weighting profiles, which I put in osmprob::weightingProfiles. You can try it with

devtools::load_all ()
q <- osmdata::opq (bbox = c (11.58, 48.14, 11.585, 48.145))
dat <- osmdata::osmdata_sf (q, "tests/osm-ways-munich.osm")
graph <- osmlines_as_network (dat, profileName = "horse")

graph now has a column d_weighted with the weighted distance.

[karpfen]

I fiddled around some more, trying to figure out what's wrong with the distance estimation. No results so far, but looking at the distance value, I saw that it's in the ballpark of the sum of all edge ids.

devtools::load_all (export_all = FALSE)
graph <- readRDS ("tests/compact-ways-munich.rds")
start_pt <- c (11.58, 48.14)
end_pt <- c (11.585, 48.145)
pts <- select_vertices_by_coordinates (graph, start_pt, end_pt)
route_start <- pts[1]
route_end <- pts [2]
prob <- get_probability (graph, route_start, route_end, eta = 1)
path <- get_shortest_path (graph, route_start, route_end)

path$distance # 0.7924514 (unit: km)
prob$distance # 584169
sum (prob$probability$edge_id) # 569309

also note that prob$distance is an integer and no float. Could it be that this is just a simple column mix up? I tried to figure it out, but r_router_prob is an enigma to us mere mortals.

[karpfen]

Actually, scratch that! I ran some more examples today and they all came out just fine.

devtools::load_all (export_all = FALSE)
start_pt <- c (10.5350035429,47.6034937672)
end_pt <- c (10.5384796858,47.6071612749)

graph <- download_graph (start_pt, end_pt)
pts <- select_vertices_by_coordinates (graph, start_pt, end_pt)
route_start <- pts[1]
route_end <- pts [2]
path <- get_shortest_path (graph, route_start, route_end)
path$distance # 0.1426759

prob <- get_probability (graph, route_start, route_end, eta = 1.0)
prob$distance # 0.191191
prob <- get_probability (graph, route_start, route_end, eta = 0.2)
prob$distance # 0.2769817
prob <- get_probability (graph, route_start, route_end, eta = 0.1)
prob$distance # 0.3704895

Maybe it was just some error in the graph from the first example or so.

[mpadge]

distance calculation generally works, so closing this and removing remaining issue to #30