Add telecommute status model #935
Conversation
This reverts commit 54bc4cc.
There was a problem hiding this comment.
Pull Request Overview
This PR implements a telecommute status model to predict whether workers telecommute on the simulation day. The model extends existing ActivitySim telecommute representation by adding a new binary model that determines daily telecommute behavior based on worker characteristics and telecommute frequency.
Key changes include:
- Implementation of a new telecommute status model with configurable probability-based logic
- Addition of documentation explaining model purpose and integration considerations
- Support for estimation workflow integration
Reviewed Changes
Copilot reviewed 4 out of 4 changed files in this pull request and generated 2 comments.
| File | Description |
|---|---|
docs/dev-guide/components/telecommute_status.md |
Comprehensive documentation for the telecommute status model component |
activitysim/estimation/larch/simple_simulate.py |
Estimation support function for telecommute status model |
activitysim/abm/models/telecommute_status.py |
Core implementation of the telecommute status prediction model |
activitysim/abm/models/__init__.py |
Module import registration for the new telecommute status component |
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
There was a problem hiding this comment.
This model would be included in our CI test system through the SANDAG example model configs. However, the currently open PR (ActivitySim/sandag-abm3-example#30) does more than just add telecommute status configs. I think the configs in that PR are out of date with what is desired for pulling in this code considering it has changes in CDAP and non-mandatory tour frequency. Do we want to just create a new PR into the sandag-abm3-example for just the telecommute_status configs and leave the rest open for future telecommute enhancements?
| @@ -49,4 +49,5 @@ | |||
| vehicle_allocation, | |||
| vehicle_type_choice, | |||
| work_from_home, | |||
| telecommute_status, | |||
There was a problem hiding this comment.
while not actually important or necessary, it would please me to have this moved up into alphabetical order like the rest of the models :)
| """ | ||
|
|
||
| preprocessor: PreprocessorSettings | None = None | ||
| """Setting for the preprocessor.""" |
There was a problem hiding this comment.
Can remove the preprocessor since it is no longer needed after merging in #957 (exists as part of the LogitComponentSettings now)
| # - preprocessor | ||
| preprocessor_settings = model_settings.preprocessor | ||
| if preprocessor_settings: | ||
| locals_d = {} | ||
| if constants is not None: | ||
| locals_d.update(constants) | ||
|
|
||
| expressions.assign_columns( | ||
| state, | ||
| df=choosers, | ||
| model_settings=preprocessor_settings, | ||
| locals_dict=locals_d, | ||
| trace_label=trace_label, | ||
| ) |
There was a problem hiding this comment.
| # - preprocessor | |
| preprocessor_settings = model_settings.preprocessor | |
| if preprocessor_settings: | |
| locals_d = {} | |
| if constants is not None: | |
| locals_d.update(constants) | |
| expressions.assign_columns( | |
| state, | |
| df=choosers, | |
| model_settings=preprocessor_settings, | |
| locals_dict=locals_d, | |
| trace_label=trace_label, | |
| ) | |
| expressions.annotate_preprocessors( | |
| state, | |
| df=choosers, | |
| locals_dict={} if constants is None else constants, | |
| skims=None, | |
| model_settings=model_settings, | |
| trace_label=trace_label, | |
| ) | |
Update with the new preprocessor function call
| some suggest it should be applied after the CDAP model only to workers who have work activities | ||
| during the day regardless of in-home or out-of-home (which requires change in CDAP definition). | ||
| The Consortium is currently engaged in an explicit telecommute design task as part of Phase 9B, | ||
| out of which more guidance on the model sequence will be established. |
There was a problem hiding this comment.
Can this text be updated to reflect the most recent design decisions around telecommuting?
This PR adds the software infrastructure to implement a telecommute status model that explicitly identifies workers who telecommute on the simulation day. This work is funded under Phase 9B.
The software adds a
is_telecommutingattribute to the persons table. A simple implementation of the telecommute status model can be based on the worker's telecommute frequency. For example, if a worker telecommutes 4 days a week, then there is a 80% probability for them to telecommute on the simulation day. The telecommute status model software can accommodate more complex model forms if needed. An example telecommute status model specification can be found in ActivitySim/sandag-abm3-example#30In addition to identifying workers who telecommute on the simulation day, we would also like to explicitly simulate their work activities on the day and let that influence non-work activities. The original scope includes overwriting the simulation day work location to home for workers who telecommute on the day, and using the existing ActivitySim mandatory tour generation submodels to simulate in-home work activities. Upon further discussion with the Consortium, we decided to not fully implement the original scope. Therefore this PR does not include the software changes to overwrite the simulation day work location to home.
There are other options to explicitly simulate telecommute behaviors on the simulation day. The Consortium is currently engaged in an explicit telecommute design task to decide the model and software design.
Related issue: #736