Various winter improvements #242

Mind-the-Cap · 2025-12-22T11:13:02Z

New classes Parameter and ParameterSet to better manage parameters of the model

Normally ensures compatibility with existing scripts, but I'm not merging, as it changes the way parameters are I want to have a discussion on that before merging!

Improvements for quickstart
Move old classes to a legacy folder
New default parameters for modes and motives (to improve with future calibration)
Add a class MultiSeedPopulationTrips to get results using multiple seeds while not complicating PopulationTrips too much
Add WorkHomeFlows_fr class to compare with INSEE flows
Use this INSEE data to create a SSI function (Sørensen similarity index)

New classes Parameter and ParameterSet to better manager parameters of the model

* bug quickfix for small radiuses in get_prominent_cities * better legend in plot_od_flows (still have to improve it) * ability to save OD flows as an image (in local folder for now, to improve)

MultiSeedPopulationTrips with some test code

codecov · 2025-12-31T16:41:49Z

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 70.42%. Comparing base (c8c3ac3) to head (cfd4891).

Additional details and impacted files

@@           Coverage Diff           @@
##             main     #242   +/-   ##
=======================================
  Coverage   70.42%   70.42%           
=======================================
  Files          56       56           
  Lines        2424     2424           
=======================================
  Hits         1707     1707           
  Misses        717      717

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

FlxPo · 2026-01-26T10:00:22Z

mobility/motives/work.py

This 0.9999999 default value seems too high to me ? 0.99986 comes from the radiation model with selection paper from Simini (https://arxiv.org/pdf/1206.4359).

That was the value I found using tests, but we will need to run proper tests soon anyway on diverse territories, so we can discuss best value later!

OK we will have to validate this, because 0.9999999 probably means we remove all short ODs, because they will not be sampled.

FlxPo · 2026-01-26T10:05:04Z

mobility/choice_models/population_trips_multi.py

Could we make a single PopulationTrips class with the option to have multiple seeds ? We probably should remove all the main as this should not run as a script anyway ?

Happy to discuss it! I hesitated for a long time indeed, here's why having a proper class was my option:

We still hope to have good results with using a single seed, ie not only use cases will imply multiple seeds

Using multiple seeds would mean we have to rewrite every function in PopulationTrips, which is a big change and creates a long file

On the opposite, by having a MultiSeedPopulationTrips, we can have a clear class for that purpose, which uses PopulationsTrips as it is and keeps it stable. That also opens the door to MultiParamsPopulationTrips (one single seed but slightly different value of costs of time for example), which could be useful for future research
(Once we agree on the solution, we can obviously remove the "if name == "main":" part)

My thinking was that single seed is just a special case of multiple seed ?

I wonder if we could modify the Asset class to handle parametric inputs automatically, so we don't need a special class every time.

FlxPo · 2026-01-26T10:07:08Z

examples/quickstart-fr.py

k_mode_sequences=42 ? What's the idea here ?

That was a test, we can switch back to a lower value (but we'll need to test the sensibility to that soon)

FlxPo · 2026-01-26T10:17:32Z

mobility/parsers/work_home_flows.py

I think WorkplaceByResidenceCounts would be a better name ? INSEE does not really give flows. Maybe remove the fr suffix, create a generic class and then have methods for each country / source ? Like in https://github.com/mobility-team/mobility/blob/main/mobility/population.py. We should be able to use swiss Statent data.

We should also avoid the name == "main" block as this should not run as a script.

INSEE urls are unstable, so we should copy the files on data.gouv.fr first.

Agree with the comments, I will do these changes asap.

FlxPo · 2026-01-26T10:21:28Z

mobility/transport_modes/bicycle/bicycle_mode.py

25 seems huge, except if you want to exclude bicycle as a mode ? It means that cycling will get a 0 % probability of being selected for nearly all comparisons with the other modes ?

Bicycle was really overrepresented otherwise. We'll need to investigate that in the near future!

FlxPo · 2026-01-26T15:53:27Z

mobility/model_parameters.py

Good first version ! I think we can avoid the post_init / dict serialization. Should I take this in a separate PR ? I think it's substantial enough.

A good ref for us : https://docs.astropy.org/en/stable/_modules/astropy/modeling/parameters.html#Parameter

Mind-the-Cap added 3 commits November 13, 2025 23:18

ongoing tests

ed9068d

ongoing tests

2467fa0

Parameter class

76e88e5

New classes Parameter and ParameterSet to better manager parameters of the model

Mind-the-Cap linked an issue Dec 22, 2025 that may be closed by this pull request

Organiser les paramètres du modèle #241

Open

Mind-the-Cap changed the title ~~Parameter class~~ Improvements for quickstart and parameters Dec 22, 2025

correct method name

a3d515f

Mind-the-Cap force-pushed the new-parameters branch from c321a2a to a3d515f Compare December 22, 2025 13:31

other improvements and experiments

3c2eea8

Mind-the-Cap force-pushed the new-parameters branch from be366a3 to 3c2eea8 Compare December 22, 2025 14:31

Mind-the-Cap added 2 commits December 22, 2025 16:06

exclude parameters config from comparison

36845b4

test

72dd752

Mind-the-Cap force-pushed the new-parameters branch from c4fd94c to 72dd752 Compare December 22, 2025 15:30

other test

f9e961b

Mind-the-Cap force-pushed the new-parameters branch from 97fb3de to f9e961b Compare December 22, 2025 18:03

access parameters via a dict

1ae982f

Mind-the-Cap force-pushed the new-parameters branch from 5d7f05b to 1ae982f Compare December 22, 2025 18:30

move old classes to legacy folder

b245677

Mind-the-Cap force-pushed the new-parameters branch from f8d9118 to b245677 Compare December 31, 2025 16:15

Mind-the-Cap added 7 commits December 31, 2025 17:17

WorkHomeFlows_fr class to compare with INSEE flows

5619e90

new default parameters

2aa54ee

minor improvements on PopulationTrips

ae60101

* bug quickfix for small radiuses in get_prominent_cities * better legend in plot_od_flows (still have to improve it) * ability to save OD flows as an image (in local folder for now, to improve)

MultiSeedPopulationTrips (in progress)

8197f95

MultiSeedPopulationTrips with some test code

SSI in results and use of that for radius sensitivity analysis

c9c48f5

last use of parameters.something ?

4abff2e

Merge branch 'main' into new-parameters

cfd4891

Mind-the-Cap changed the title ~~Improvements for quickstart and parameters~~ Various winter improvements Dec 31, 2025

Mind-the-Cap mentioned this pull request Jan 5, 2026

[US] Pouvoir comparer les résultats de la modélisation avec des données de référence #153

Open

3 tasks

FlxPo reviewed Jan 26, 2026

View reviewed changes

Various winter improvements #242

Are you sure you want to change the base?

Various winter improvements #242

Uh oh!

Conversation

Mind-the-Cap commented Dec 22, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

codecov bot commented Dec 31, 2025

Codecov Report

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Mind-the-Cap commented Dec 22, 2025 •

edited

Loading