janowicz
commented
6 years ago This is ready for review / testing.
Summary of changes:
- HLCM specs in /configs/hlcm/regional
- ELCM specs in /configs/elcm/regional
- Updated the lists of model configs to use in /configs/yaml_configs.yaml
- In models.py, update the location choice model registration code to apply the regional specifications on a by-large-area basis. Also added an optional orca step to calculate diagnostic indicators (aggregate summaries of share of emp sector in each general type building, specific summaries of counts of jobs in each of the "case study" buildings)
- Added general type dummy variables in variables/variables_buildings.py
This PR represents the regional ELCM/HLCM model specs. Each submodel (ELCM by sector_id, HLCM by income quartile) will be estimated based on a regional sample of agents rather than by large-area. Evaluation of the specifications includes consideration of summed-probability-by-building-type plots to check for the expected relationship between the submodel (e.g. healthcare jobs) and building type (e.g. medical structures). Any specification for which the estimation process emits a non-convergence warning is rejected.