MAINT: statespace cleanup, alphabetically

[jbrockmendel]

@ChadFulton this is the second of three statespace branches that have been sitting ready to go for a while, then I'll stop creating more work for you.

[coveralls]

Coverage increased (+0.01%) to 84.839% when pulling 3a99c6f on jbrockmendel:stests6 into 68d5436 on statsmodels:master.

[codecov]

Codecov Report

Merging #5779 into master will increase coverage by <.01%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master    #5779      +/-   ##
==========================================
+ Coverage    82.3%   82.31%   +<.01%     
==========================================
  Files         593      593              
  Lines       93728    93761      +33     
  Branches    10356    10356              
==========================================
+ Hits        77146    77179      +33     
  Misses      14185    14185              
  Partials     2397     2397

Impacted Files	Coverage Δ
statsmodels/tsa/statespace/tests/test_kalman.py	95.36% <100%> (ø)	⬆️
...models/tsa/statespace/tests/test_dynamic_factor.py	97.71% <100%> (+0.06%)	⬆️
...tsmodels/tsa/statespace/tests/test_concentrated.py	100% <100%> (ø)	⬆️
statsmodels/tsa/statespace/tests/test_collapsed.py	96.55% <100%> (+0.01%)	⬆️
...els/tsa/statespace/tests/test_impulse_responses.py	100% <100%> (ø)	⬆️
...models/tsa/statespace/tests/test_initialization.py	100% <100%> (ø)	⬆️
...a/statespace/tests/test_exact_diffuse_filtering.py	98.16% <100%> (+0.01%)	⬆️

---

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 68d5436...3a99c6f. Read the comment docs.

[codecov]

Codecov Report

Merging #5779 into master will increase coverage by <.01%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master    #5779      +/-   ##
==========================================
+ Coverage    82.3%   82.31%   +<.01%     
==========================================
  Files         593      593              
  Lines       93728    93761      +33     
  Branches    10356    10356              
==========================================
+ Hits        77146    77179      +33     
  Misses      14185    14185              
  Partials     2397     2397

Impacted Files	Coverage Δ
statsmodels/tsa/statespace/tests/test_kalman.py	95.36% <100%> (ø)	⬆️
...models/tsa/statespace/tests/test_dynamic_factor.py	97.71% <100%> (+0.06%)	⬆️
...tsmodels/tsa/statespace/tests/test_concentrated.py	100% <100%> (ø)	⬆️
statsmodels/tsa/statespace/tests/test_collapsed.py	96.55% <100%> (+0.01%)	⬆️
...els/tsa/statespace/tests/test_impulse_responses.py	100% <100%> (ø)	⬆️
...models/tsa/statespace/tests/test_initialization.py	100% <100%> (ø)	⬆️
...a/statespace/tests/test_exact_diffuse_filtering.py	98.16% <100%> (+0.01%)	⬆️

---

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 68d5436...ca52a73. Read the comment docs.

[jbrockmendel]

@bashtage additional failure mode ssl.SSLError for get_rdataset.

[bashtage]

TBH I don't really see the point of switching existing assert_ to assert or assert_equal to assert A == B. It creates a lot of churn and there is effectively 0 gain since NumPy will never drop these functions.

I'm more positive about pytest.raises since this produces more natural function calls.

[jbrockmendel]

I didn't count, but many of the asserts are for assert foo is None which is a strict (albeit small) improvement over an equality check. Many others are for regex boolean checks which are really non-idiomatic in the status quo. For the other cases, I'll keep your comment in mind for future branches.

[ChadFulton]

@jbrockmendel sorry about the delay, but TBH you make it somewhat difficult to consider whether or not to accept some of these PRs because you mix things like easy-to-commit pep8 improvements with opinion-based code changes like using the pytest.raises context.

I'm undecided here.

[jbrockmendel]

You do you buddy. I just wanted to get this up so it doesn’t rot during my next leave of absence.

[ChadFulton]

Just so you know where I'm coming from:

• I would have accepted a pep8 improvement patch and / or one that replaced assert_equal with assert or assert_
• I would have rejected a patch that removed assert_raises.

Since you don't seem to be in a compromising mood, we'll close this for now.

[jbrockmendel]

Totally happy to compromise, will put together flake8-only version.

Out of curiosity, any idea why the temp_filename fixture in test_save is needed? Doesnt NamedTemporaryFile do exactly the same thing?

[jbrockmendel]

in test_mlemodel.test_score_analytic_ar1 approx_fd_centered is defined but never used. im going to make the opinionated edit and decide it was intended to be part of the assertion in the following line