Fix bug on force plots to show feature values for specific rows in training data

[peterataylor]

…Fixes bug where only first row feature values are show in all row plots.

Pull Request Description

Fixes bug as described in #3045.

---

After creating the pull request: in order to pass the release_notes_updated check you will need to update the "Future Release" section of docs/source/release_notes.rst to include this pull request by adding :pr:123.

[CLAassistant]

All committers have signed the CLA.

[codecov]

Codecov Report

Merging #3044 (917f078) into main (2598cfa) will increase coverage by 0.1%.
The diff coverage is 100.0%.

@@           Coverage Diff           @@
##            main   #3044     +/-   ##
=======================================
+ Coverage   99.8%   99.8%   +0.1%     
=======================================
  Files        313     313             
  Lines      30567   30579     +12     
=======================================
+ Hits       30477   30489     +12     
  Misses        90      90             

Impacted Files	Coverage Δ
evalml/model_understanding/force_plots.py	100.0% <100.0%> (ø)
.../prediction_explanations_tests/test_force_plots.py	100.0% <100.0%> (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 2598cfa...917f078. Read the comment docs.

[freddyaboulton]

@peterataylor Do you need help getting all the checks to pass?

[peterataylor]

@peterataylor Do you need help getting all the checks to pass?

Thanks for the offer @freddyaboulton ! Having a quick look I should be able to do the release notes. But any guidance on the lint, code coverage and dependency changes would be great. But I will look over the contrib guide again and check the approach.

[freddyaboulton]

@peterataylor The "Detect dependency changes" should be fixed as soon as you merge in the latest main branch.

If you have the development dependencies installed, you should be able to do the following from the top level of the repo:

make lint-fix
make lint

If make lint still fails, the errors should hopefully be clear.

If you do not have make installed you can look at the Makefile to see what those commands mean and run then yourself.

Let me know if you have any other questions!

[freddyaboulton]

@peterataylor This looks good to me! Thanks for the contribution. I think the one thing that's missing is maybe a unit test to make sure we don't commit the same mistake again in the future. Maybe do one with and without a custom index on training_data?

[peterataylor]

Thanks @freddyaboulton . I've added some tests to test_force_plots that use the existing parameter sets (rows 0 and 0-4). It required inspection of the returned plot object to ensure the feature values matched those that correspond to the rows_to_explain. I confirmed the tests failed with previous version and pass now. You will probably want to review though. Thanks!

[freddyaboulton]

@peterataylor This looks good to me! Thank you for the contribution.