dodgr flows

library (dodgr)

The dodgrpackage includes three functions for allocating and aggregating flows throughout network, based on defined properties of a set of origin and destination points. The three primary functions for flows are dodgr_flows_aggregate(), dodgr_flows_disperse(), and dodgr_flows_si(), each of which is now described in detail.

1 Flow Aggregation

The first of the above functions aggregates ''flows'' throughout a network from a set of origin (from) and destination (to) points. Flows commonly arise in origin-destination matrices used in transport studies, but may be any kind of generic flows on graphs. A flow matrix specifies the flow between each pair of origin and destination points, and the dodgr_flows_aggregate() function aggregates all of these flows throughout a network and assigns a resultant aggregate flow to each edge.

For a set of nf points of origin and nt points of destination, flows are defined by a simple nf-by-nt matrix of values, as in the following code:

graph <- weight_streetnet (hampi, wt_profile = "foot")
from <- sample (graph$from_id, size = 10)
to <- sample (graph$to_id, size = 10)
flows <- matrix (10 * runif (length (from) * length (to)),
                 nrow = length (from))

This flows matrix is then submitted to dodgr_flows_aggregate(), which simply appends an additional column of flows to the submitted graph:

graph_f <- dodgr_flows_aggregate (graph, from = from, to = to, flows = flows)
head (graph_f)

Most flows are zero because they have only been calculated between very few points in the graph.

summary (graph_f$flow)

2 Flow Dispersal

The second function, dodgr_flows_disperse(), uses only a vector a origin (from) points, and aggregates flows as they disperse throughout the network according to a simple exponential model. In place of the matrix of flows required by dodgr_flows_aggregate(), dispersal requires an equivalent vector of densities dispersing from all origin (from) points. This is illustrated in the following code, using the same graph as the previous example.

dens <- rep (1, length (from)) # uniform densities
graph_f <- dodgr_flows_disperse (graph, from = from, dens = dens)
summary (graph_f$flow)

3 Merging directed flows

Note that flows from both dodgr_flows_aggregate() and dodgr_flows_disperse() are directed, so the flow from 'A' to 'B' will not necessarily equal the flow from 'B' to 'A'. It is often desirable to aggregate flows in an undirected manner, for example for visualisations where plotting pairs of directed flows between each edge if often not feasible for large graphs. Directed flows can be aggregated to equivalent undirected flows with the merge_directed_graph() function:

graph_undir <- merge_directed_graph (graph_f)

Resultant graphs produced by merge_directed_graph() only include those edges having non-zero flows, and so:

nrow (graph_f); nrow (graph_undir) # the latter is much smaller

The resultant graph can readily be merged with the original graph to regain the original data on vertex coordinates through

graph <- graph [graph_undir$edge_id, ]
graph$flow <- graph_undir$flow

This graph may then be used to visualise flows with the dodgr_flowmap() function:

graph_f <- graph_f [graph_f$flow > 0, ]
dodgr_flowmap (graph_f, linescale = 5)

Try the dodgr package in your browser

Any scripts or data that you put into this service are public.

dodgr documentation built on Aug. 8, 2021, 1:06 a.m.