search

ATFutures / upthat

Urban planning and transport health assessment tool
4 stars 1 forks source link

Release new flow layers in upthat releases #21

Closed Robinlovelace closed 4 years ago

Robinlovelace commented 4 years ago
mpadge commented 4 years ago

First full cuts for Accra and Kathmandu now uploaded. All works perfectly on my local instance, so hopefully should auto-work on public server. We need to also think about how to incorporate tests with data releases to ensure any changes there will actually work? Anyway, it should just work with demo data for Kathmandu only until release data have been downloaded to local inst, after which Kathmandu should be automatically updated to the full analysis. There is thus no need for "Demo" in the above list, yet we still retain the "demo" functionality in the absence of any additional data, which is (i think) nice.

Edit: This issue is not quite ready to close, because i still need to properly scale Kathmandu to local data, and also to include NYC. But most importantly, i need address #22 first, which includes embedding a few scaling parameters as attributes of each net-<city>.Rds file. I've done that for Accra and Kathmandu, but not yet enabled the use of those parameters for scenario switching.

Robinlovelace commented 4 years ago

Great work @mpadge, should have checked before chatting, will test here asap.

Robinlovelace commented 4 years ago

Heads-up @mpadge I just tested the new version of the data on the current branch and works good, Accra shown below.

image

Robinlovelace commented 4 years ago

Just testing with latest PR. One issue: the lines still rather faint and thin. But great to see it working well for Accra :+1:

image

Robinlovelace commented 4 years ago

@mpadge could you update the demo data in the inst folder?

Robinlovelace commented 4 years ago

Also looks good for Kathmandu as shown below. But it's not clear what the units are: I would expect more than 700 pedestrians to pass the busiest streets in the city (and more than 255 in Accra).

image

mpadge commented 4 years ago

@Robinlovelace We need to decide here which layers we are going to keep, and which not. As it really is MVP, my first suggestion would be just the main 2. Thoughts?

Robinlovelace commented 4 years ago

We said we'd release additional cities compared with Phase II. Can you share the code used to build the latest flow layers? From that it should be easy to build data for Bristol and NY (at a minimum).

mpadge commented 4 years ago

NY can do, but Bristol is actually really tricky without knowing aggregate passenger numbers at Temple Meads (and other suburban stations). Bristol's PT is too multi-modal for NY model to translate directly absent a bit of additional calibration info

Robinlovelace commented 4 years ago

Doesn't need to be accurate, just a proof of concept.