mm80843 / paris_maps

MyWay in Paris

App running at https://mparismap.herokuapp.com/ .

Content

• Streamlit app folder
• 00.PreparingData.ipynb (pre processing from opendatasets)
• 01.Stats.ipynb: initial insights (to be completed)
• 02.PreparingRouteFinder.ipynb: draft router finder
• 03.CastingShadows.ipynb: running tests with shadows mapping

Done

• Zoning on arrondissements 18, 19, 20, 10, 11
• Been getting all routes from within OSM
• Preliminary mapping of best route based on trees
• Trees and lighting are put in
• Add shadows

Other data layers?

• Use distance to fountains, parcs, green spaces
  • What other layers to use?
• Develop something from building footprint (in data/BoI_simple.pkl)

Images

• Example of the network

• Merging all buildings in one

• Details of streets

Learn and try

The data processing API

1. Find

• Should get: X,Y start, X,Y end + set of preferences ("weights")
• Should receive, waypoints (nodes) (ordered list of XYs), and layers of interest close to the path (XYs of trees, lights, .. )

2. Up: tells if the API is up
3. Discover: sends end start users, settings - cache available

Goodies, API side

Caching

• Define
  • Defining them: So to have a couple of from-to couples, and personas (ie prepared sets of weights)
  • Store results in dataframe
• Run route finders and cache results
  • Start
  • End
  • Weights
  • Returns: json with waypoints, trees, lights, parcs, shadows

Visualisation

• Streamlit:

  • inputs
  • map
  • description of the map on the right

• Personas + "you"

  • Luc: home->work (day), then work->home night (shadows, light)
  • Anna: montmartre->b&b in 19th : trees and lights
  • Laure: South of 12th -> la villette

Logs and ideas

D1: 20210723 Starting

• Been preparing the code as in a python standalone module.
  • Code as usual is in the "module" folder, here "pypama" (python paris maps)
  • Version is generated from the module itself, as in the pypama/version.py file
  • Adding binaries (the data pickles as part of the module in MANIFEST.in)
  • Added a setup.py file to generate the module
  • gitignore file prepared to exclude all "useless" bits from the repo
  • Added some support scripts (build and install to generate the module)
• Discovered how to generate a requirements.txt file for projects. Useful, and added in the [createReqs.sh](createReqs.sh) script.
• From there, preparing a "TestingModule" notebook to try out changes in the module as the same time it's tested. The %load_ext autoreload %autoreload 2 magic comes in handy to avoid reloading everything at the same time.

D2: 20210724 Going till the app

• Release e74e19ff
  • Prepared the 99.app.py launched by streamlit run 99.app.py. Opens up an app window to play around with weigths.
  • Text inputs ready for the search
  • Formula in calculatePath still needs to be finetuned.
  • No API yet, just to play around with some of the layers. Caching still to be optimized =)
  • v0.0.1 with workable version is released!

D3: 20210803 Publishing an API

• FastAPI-based API branch pushed here

  • Running with uvicorn main:app --reload
  • Fast API, once launched, can be tested through http://127.0.0.1:8000/docs#/
  • Getting starting and ending points, returning the optimized path as well as shortest path
  • Result looks good:

• API-dependant streamlit app branch pushed here

  • Badly coded, list of waypoints sent as strings and then parsed on client end, but working.
  • Works well, needs to setup the server URL, but appears to be working smoothly

• @todo: pull the branches on mm80843, connecting two free servers from heroku

Changelog

v0.0.2

• Ongoing

v0.0.1

• First workable version is released!
• Contains building footprints, shadows, lights, trees, green spaces, water points..
• Route finding is done!