[WIP] Test Coverage for Line Methods, Changes to Line Methods

[marty-larocque]

Primarily, this PR addresses issue #186 (test coverage). It adds tests for base.assert_line() and base.assert_lines_of_type(). These tests should also provide coverage for base.get_slope_yintercept().

While working on these tests, a number of other issues were encountered. This PR should also address those:

• X-limits issue with base.assert_line() and seaborn.regplot() #235 and Line transform issue with base.assert_line() and seaborn.regplot() #236 (Issues with using seaborn)
• Change to line_types parameter of base.assert_lines_of_type() #237 (A small improvement)
• ValueError when calling assert_lines_of_type() with line_types='regression' #238 (A case that produces an error)
• Issue checking x-limits of lines without additional plotted data. #262 (An edge case that produces an error)

[marty-larocque]

Currently, this PR is not ready to be merged for a few reasons:

• The way that X-limits issue with base.assert_line() and seaborn.regplot() #235 is currently addressed in this PR is just a quick and dirty fix. We should decide on our approach to this issue before this PR is merged.
• The way that ValueError when calling assert_lines_of_type() with line_types='regression' #238 is currently addressed just suppresses the error. It is not a reliable fix. assert_lines_of_type() likely needs to be completely refactored. We should also decide on our approach here.
• There are currently no docstrings for the tests.
• The tests require seaborn, which is not included in the requirements.

[marty-larocque]

This PR has been cleaned up a bit:

• ValueError when calling assert_lines_of_type() with line_types='regression' #238 is addressed in a better way
• Docstrings have been added to the line tests
• Seaborn has been added to the dev requirements

However:

• X-limits issue with base.assert_line() and seaborn.regplot() #235 is still unaddressed. The test test_line_type_reg_one2one is failing for this reason. I believe that Nathan is working on a fix for this issue.
• Issue checking x-limits of lines without additional plotted data. #262 is an issue I discovered recently, and is not yet addressed. The test test_line_type_one2one is failing for this reason. I believe that this is because the test here represents an edge case in which no data is plotted, but a line is plotted. We should either change the test to represent a more common case, or we should address this issue.

Finally:

• Line transform issue with base.assert_line() and seaborn.regplot() #236 is still addressed by this PR
• Change to line_types parameter of base.assert_lines_of_type() #237 is still addressed by this PR

We should wait until Nathan's PR addressing #235 is merged before we consider merging this PR. We should also address #262, one way or another, before we merge this PR.

[lwasser]

lets:

1. Remove xtime as a parameter
2. Let's make the data extent (boolean) parameter to make this test OPTIONAL

[lwasser]

Here - let's address the issue of making the xlims data test a parameter (boolean)
please also

1. Update from master (so the pr passes)
2. Add a change log entry
3. Add any comments at the end for anything that hasn't been complete!

[codecov]

Codecov Report

Merging #239 into master will increase coverage by 1.46%.
The diff coverage is 87.32%.

@@            Coverage Diff             @@
##           master     #239      +/-   ##
==========================================
+ Coverage   79.95%   81.42%   +1.46%     
==========================================
  Files          19       20       +1     
  Lines        1831     1884      +53     
==========================================
+ Hits         1464     1534      +70     
+ Misses        367      350      -17     

Impacted Files	Coverage Δ
matplotcheck/base.py	93.52% <65.38%> (+5.81%)	⬆️
matplotcheck/tests/test_base_lines.py	100.00% <100.00%> (ø)

---

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 eab6e72...ae7c000. Read the comment docs.

[marty-larocque]

@lwasser This PR should be ready to be merged. Just note that #262 still needs to be addressed. That is not anything urgent, and it's a fairly rare edge case, but it should be addressed at some point.

[lwasser]

TODO's here:

• is there another way in python to calculate a regression line? if so - should we add a test for that approach to ensure this works
• let's change regression option to linear-regression

think about edge cases...

[nkorinek]

@lwasser this can be closed due to this pr: #274

[lwasser]

ok thanks @nkorinek closing!!