Interdependent recovery of residential buildings and households

[ylyangtw]

This PR is to support Lisa Wang's new analysis Interdependent recovery of residential buildings and households.

Two analyses have changes:

1. Residential building recovery
   • Add 2 additional output datasets, which are the same as commercial building recovery
2. Housing recovery sequential
   • Major idea is that zones can be generated based on either social vulnerability or household income.
   • Change sv_generator from hardcoded to 2 input datasets zone_def_sv and zone_dev_hhinc. To make sure analysis runs without inputs, default zone_def_sv and zone_dev_hhinc are provided.
   • Add one function to generate zones based on household income, and modify the model accordingly.

For testing, please run
test_residentialbuildingrecovery.py
test_housingrecoverysequential_sv.py
test_housingrecoverysequential_hhinc.py

[longshuicy]

Code runs well besides the minor suggestions. Approved.

[navarroc]

Can we make this a dataset on dev? This is a pretty big file and is better served from the incore-dev and referenced by ID instead.

[navarroc]

The test folder should match the analysis folder name. Since the folder is housingrecoverysequential, the test folder shouldn't be housingrecoveryserial it should be housingrecoverysequential. This looks like it was missed before. Can you rename that folder?

[navarroc]

test_housingrecoverysequential_sv.py raises a value error about hhinc, is that expected?

[navarroc]

This can probably be shortened to match the 10 char limit for dbf/shapefile. Maybe zone_indic?

[navarroc]

Do we need to open issues to add new semantic types for the outputs from residential building recovery?

[navarroc]

test_housingrecoverysequential_sv.py raises a value error about hhinc, is that expected?

@longshuicy Raising a value error if hhinc is not present prevents the analysis from running, should it do that? It looks like the population_dislocation dataset referenced in the test doesn't have that column, probably because the housing unit allocation dataset used to generate it didn't have that column. If hhinc should be there, then we should raise a value error and fix the dataset on dev. Since this isn't a test we execute as part of the pyincore testing, we could fix this later so we can get this merged.

[ylyangtw]

test_housingrecoverysequential_sv.py raises a value error about hhinc, is that expected?

@longshuicy Raising a value error if hhinc is not present prevents the analysis from running, should it do that? It looks like the population_dislocation dataset referenced in the test doesn't have that column, probably because the housing unit allocation dataset used to generate it didn't have that column. If hhinc should be there, then we should raise a value error and fix the dataset on dev. Since this isn't a test we execute as part of the pyincore testing, we could fix this later so we can get this merged.

The original population_dislocation_block (Galveston) doesn't have hhinc and it calculates the zones using social vulnerability. To support Lisa Wang's analysis, Joplin population_dislocation_block has hhinc and zones are calculated using hhinc.

So we don't need to raise a value error if hhinc is not there since hhinc is not required. It's only a condition - If hhinc is there, then we add it to pop_dis_selectors

Sorry I didn't think it clearly and test properly when I made the changes.

[ylyangtw]

Do we need to open issues to add new semantic types for the outputs from residential building recovery?

Just opened an issue #611

[ylyangtw]

@navarroc I have addressed all the comments. Thank you!