Closed anastassiavybornova closed 1 week ago
@anastassiavybornova What's the status here?
sooo. the edits to the BikeDNA
repo have been made and merged (see here: https://github.com/anerv/BikeDNA/pull/2) but with the changes introduced to core & utils in the meantime, (and also now that we don't need to copy paste code but can pip install bikedna), i wouldn't merge this one but rather redo the feature matching from scratch, i think that'll be the easiest. we can put that on the issues / to do list & assigned to me, but i think that's lower priority than the COINS stuff for now.
(i'm fine with deleting this branch but leave it up to you @jGaboardi )
(i'm fine with deleting this branch but leave it up to you @jGaboardi )
Let's leave it for now until we're 100% it will either be superseded or not needed.
adding a first draft of a feature matching workflow, adapted (no meaningful changes, just variable renaming) from my colleague Ane here: https://github.com/anerv/BikeDNA_BIG/tree/main (most of the code comes from: https://github.com/anerv/BikeDNA_BIG/blob/main/feature_matching_hpc/scripts/feature_matching.py )
featurematching_compute.ipynb
with helper functions inmatching_functions.py
runs the feature matching algorithm for a selected city - i ran it for auckland now, with a resolution of 10m (segment length), which took a couple of hoursdata/{fua_id}/evaluation/
contains all the output files from the notebookfeaturematching_plot.ipynb
contains some visualization of the feature matching resultsthere are probably some efficiency gains to be made from you digging into some of the
matching_functions.py
stuff, but at least it already works for now