Current MMR workflow seeks the ability to combine multiple overlapping/adjoining scenarios into a single merged scenario and then apply a sequence to the 'modified' scenario via sequence_plan_far().
One (of several problems) from this desired functionality is the creation of additional fake nodes. AKA - these are vertices from networks-proposed.shp, vertex_df, that do not correlate with either (1)metrics-local.csv or (2) networks-existing.shp.
Therefore, I recommend we determine authentic "fake" start nodes as those that actually intersect with networks-existing.shp. We will need to read that shapefile explicitly in NetworkPlanner.R like such: network_ext <- readOGR(dsn=base_dir, layer="networks-existing"). In essence, the assumption that all fake nodes are start nodes may not hold up in these modified scenarios.
Consequentially, we should make a new category for vertices in addition to 'connected' and 'disconnected.' Perhaps we can call them 'interconnected' because these floating vertices often occur when two separate branches interconnect.
I think at the heart of it, we would have to modify the separate_subgraphs() function. Any vertices that are not 'connected' or 'disconnected' should have an attribute 'interconnected.' 'Interconnected' nodes should always come towards the end of sequence algorithm and be given a demand or Household value that is very very small say 1.0e-3.
Right now, when I try to run the sequencer on these modified networks, I am thrown for a loop here in the create_graph() function.
Current MMR workflow seeks the ability to combine multiple overlapping/adjoining scenarios into a single merged scenario and then apply a sequence to the 'modified' scenario via
sequence_plan_far().One (of several problems) from this desired functionality is the creation of additional fake nodes. AKA - these are vertices from networks-proposed.shp,
vertex_df, that do not correlate with either (1)metrics-local.csv or (2) networks-existing.shp.Therefore, I recommend we determine authentic "fake" start nodes as those that actually intersect with networks-existing.shp. We will need to read that shapefile explicitly in NetworkPlanner.R like such:
network_ext <- readOGR(dsn=base_dir, layer="networks-existing"). In essence, the assumption that all fake nodes are start nodes may not hold up in these modified scenarios.Consequentially, we should make a new category for vertices in addition to 'connected' and 'disconnected.' Perhaps we can call them 'interconnected' because these floating vertices often occur when two separate branches interconnect. I think at the heart of it, we would have to modify the separate_subgraphs() function. Any vertices that are not 'connected' or 'disconnected' should have an attribute 'interconnected.' 'Interconnected' nodes should always come towards the end of sequence algorithm and be given a demand or Household value that is very very small say 1.0e-3.
Right now, when I try to run the sequencer on these modified networks, I am thrown for a loop here in the
create_graph()function.