Closed jdebacker closed 3 years ago
I've added an example run script to this PR and a unit test to ensure it runs for at least 5 min.
I was successfully able to solve the model under the baseline and reform used in the run_og_uk.py
script.
@rickecon @jpycroft I believe this PR is ready for your review.
Hi @jdebacker - just tried to run this but it doesn't seem to solve for me - I get a runtime error ("Steady state equilibrium not found") after about 2 minutes. Could this be a dependency issue, or something wrong in my setup?
@nikhilwoodruff Thanks for testing this out!
Your traceback doesn't sound like a dependency issue since things got kicked off and ran. Unfortunately, I cannot replicate your result.
It's possible this is related to a difference in dependent package versions. Can you post the results of a conda list
when you are in the oguk-calibrate-dev
environment?
Also, what OS are you running?
Merging #19 (8c25289) into main (961c648) will increase coverage by
7.11%
. The diff coverage is67.64%
.
@@ Coverage Diff @@
## main #19 +/- ##
==========================================
+ Coverage 51.63% 58.74% +7.11%
==========================================
Files 5 6 +1
Lines 736 766 +30
==========================================
+ Hits 380 450 +70
+ Misses 356 316 -40
Flag | Coverage Δ | |
---|---|---|
unittests | 58.74% <67.64%> (+7.11%) |
:arrow_up: |
Flags with carried forward coverage won't be shown. Click here to find out more.
Impacted Files | Coverage Δ | |
---|---|---|
og_uk_calibrate/tests/test_get_micro_data.py | 85.29% <ø> (-0.43%) |
:arrow_down: |
og_uk_calibrate/txfunc.py | 53.15% <ø> (+10.50%) |
:arrow_up: |
og_uk_calibrate/tests/test_run_example.py | 62.06% <62.06%> (ø) |
|
og_uk_calibrate/get_micro_data.py | 61.36% <100.00%> (+4.22%) |
:arrow_up: |
Continue to review full report at Codecov.
Legend - Click here to learn more
Δ = absolute <relative> (impact)
,ø = not affected
,? = missing data
Powered by Codecov. Last update 961c648...8c25289. Read the comment docs.
The only thing I can think I've done differently in the setup is that I had to download the OG-USA source code from GitHub and install that, commenting out the OGUSA requirement in the environment.yml file - git refused to clone the repo, kept running into 'remote hung up unexpectedly' errors. But since both pointing at the latest version I assumed it would be fine. My conda list output is here, and the output from run_og_uk.py is here if that helps? I'm on Ubuntu. Thanks!
@nikhilwoodruff @jpycroft Note that I just added the test of the run_og_uk.py
script to the CI tests run on GH Actions. The test passes, which means that all dependencies were installed fine on the machines used by GH Actions. I don't know if looking through the log files from these runs would be helpful.
Thanks @jdebacker - I've still not figured out what the issue is on my end, but it's good to know it runs fine in the cloud. Out of interest are you running on the actual FRS microdata (the GH action runs on the toy dataset, and I'm running on the actual 2018 FRS)?
@nikhilwoodruff Maybe I've never run on the full FRS. I've only done what you see in this PR. I think that is all the model is set up to do. The relevant line is here. Please feel free to suggest and edit to this to use the full data (if it is available to the public).
This PR makes some updates to the
Calibration
class (commenting out attributes not yet created) and toget_micro_data.py
(to choose the proper weight from the FRS data and to set 2021 as the default start and end year).