First draft of relaxed time periods #59
Conversation
… into feat/relaxed_timeperiods
There was a problem hiding this comment.
Looks good! However, I don't like the idea of including distance matrix drop to this update. Archipelago areas and ferries are such small portion of total traffic that other development work should be prioritized. On quick look there is < 0,2 % of population living in these areas and we many many open issues with the 80 percent.
Maybe we should also test the run with off peak period before merging? Therefore I didn't approve changes yet.
I restored distance matrices for walk and bike. To be able to run the model with more LOS matrices (evening added) it was necessary to remove distance at least for the modes that do not use it in demand calculation. |
… into feat/relaxed_timeperiods # Conflicts: # Scripts/tests/test_data/Network/netfield_links_test.txt # Scripts/tests/test_data/Network/netfield_transit_lines_test.txt
Scripts/tests/test_data/Base_input_data/Matrices/uusimaa/demand_ilt.omx
Outdated
Show resolved
Hide resolved
|
Työtila matrices have suffix "it" and here it seems to be "ilt". Which is preferred? |
* Update demand parameters with evening hour shares * Add demand model parameters for new time period
… into feat/relaxed_timeperiods # Conflicts: # Scripts/datatypes/purpose.py # Scripts/tests/test_data/Network/netfield_links_test.txt # Scripts/tests/test_data/Network/netfield_transit_lines_test.txt
I will change the name to "it" everywhere, but aside from that, this PR seems to be finished. |
|
@zptro I think we'll need transit impedance matrices for |
Updated time_pt matrices in sharepoint. |
This PR enables producing evening transit LOS and using them in demand model.
As a separate step we should try model estimation without the "aht" LOS, to see if that time period can be dropped in demand calculation loop.
As suggested in this PR, we should also test replacing bike and walk distance with time and to entirely drop distance matrix from demand calculation. Mostly time is a proxy for distance, but when running the model for archipelago areas, there will be a difference in how people use ferries in walk mode.To be able to run the model with more LOS matrices (evening added) it was necessary to remove distance at least for the modes that do not use it in demand calculation.TODO: Matrices and extra attributes in EMME are currently a mess! In a later PR, I will try to refactor all of that.