Conversation
…ger throws an error
|
Thanks very much for this great contribution @mxndrwgrdnr! We'll review this soon, sorry for the delay. There is an existing earlier open PR that was reveiewed, which we kept open for a bit in the case anyone from the community wanted to review, that we'll merge in first. And then we can help reconcile with this PR. |
synthpop/recipes/starter2.py
Outdated
| self.p_pums_cols = ('serialno', 'PUMA00', 'PUMA10', 'RELP', 'AGEP', | ||
| 'ESR', 'RAC1P', 'HISP', 'SEX') | ||
| self.p_pums_cols = ('serialno', 'SPORDER', 'PUMA00', 'PUMA10', 'RELP', 'AGEP', | ||
| 'ESR', 'SCHL', 'SCH', 'JWTR', 'PERNP', 'WKHP', 'RAC1P', 'HISP', 'SEX') |
There was a problem hiding this comment.
Why are you adding these variables if they are not used in the synthesis?
|
Awesome, thanks @mxndrwgrdnr! I'll check it out today |
|
thanks !! @mxndrwgrdnr |
cvanoli
left a comment
cvanoli
left a comment
There was a problem hiding this comment.
Looks good to me @mxndrwgrdnr
synthpop/ipu/ipu.py
Outdated
|
|
||
| if iterations > max_iterations: | ||
| raise RuntimeError( | ||
| warnings.warn( |
There was a problem hiding this comment.
In a conv with @janowicz we discussed that it would be better if this ignoring of errors was made optional. Could we make an ignore parameter, where, if set to True, would raise warnings instead of errors?
| household_freqs, person_freqs, household_constraints, | ||
| person_constraints): | ||
| with pytest.raises(RuntimeError): | ||
| with pytest.warns(UserWarning): |
There was a problem hiding this comment.
Idem comment to ipu.py line263.
|
I tested the branch with a local environment in my computer and two Future Warnings keep coming up:
2 - synthpop\census_helpers.py: Line 180:
|
|
@janowicz the second FutureWarning that @cvanoli mentions isn't a big deal and should resolve itself in the future. The first FutureWarning, however, is happening during IPU when household weights that are very nearly zero are not close enough to zero to be caught by |
…rs flag to trigger the UserWarning behavior instead
added python 3.7
…parallel_py3 acs year missing
4 participants
max_iterationsis reached, it generates a warning.synthesize_all_in_parallel()to synthesizer.py for processing sub-county geographies in parallel.Could use an extra set of eyes at L264-265 in synthesizer.py. These two lines account for the fact that when processing geographies in parallel,
draw.draw_households()cannot rely on thehh_index_startargument to set thehousehold_ideither as the index of the synthetic households table or thehh_idfield of the synthetic persons table. Thus we have to adjust the households index and personshh_idfield AFTER all the tables have been generated, which is what's hapenning at L264-265.