efficiency fails to compute (global & local)

[anastassiavybornova]

for both paris and lyon, lots of warnings

Problem with efficiency_local_routed.
Problem with efficiency_global_routed.

caused by calculate_metrics https://github.com/mszell/bikenwgrowth/blob/f4bee1cd17abb36c39070dfd3bb08d47e4e8b911/code/functions.py#L998C5-L998C22 possibly empty graphs?

[Froguin99]

Also experiencing this. It appears that efficiency value is almost always calculated as zero (as EG is always zero, presumably as there is no path between the input nodes) in here:

bikenwgrowth/code/functions.py

Line 944 in 3ad32f9

def calculate_efficiency_global(G, numnodepairs = 500, normalized = True):

Not sure if that is the cause of the problem or a sympton of it. Same occurs with local routed in the following function.

I've tried looking at the input data into

bikenwgrowth/code/04_analyze_results.ipynb

Line 213 in 3ad32f9

" output_MST, cov_MST = calculate_metrics(res[\"MST\"], res[\"MST_abstract\"], G_carall, nnids, output, buffer_walk, numnodepairs, debug, True, ig.Graph(), Polygon(), False, Gexisting)\n",

to try and find a source of the issue. The inputs G_carall seems okay, it doesn't have any disconnected components so paths between all nodes should be possible. Similarly, the graphs loaded from the resultsfile look visually correct when I plotted them prior to calculcating the metrics.

Any other ideas of where I could try and debug? Still getting my head around how everything fits together 😄