Implement CounterPlots to DiCE CF generation

[rmazzine]

Dear DiCE developing team,

This PR adds to DiCE a novel way to represent counterfactual explanations. The current DiCE only shows the explanations as a list of parameters, and we know the importance of having more graphically appealing representations, especially when creating explanations.

Therefore, we've developed a new package to generate metrics and charts that augment the informative value of counterfactual explanations, CounterPlots (https://github.com/ADMAntwerp/CounterPlots), the preprint article with full information can be found in this link: https://arxiv.org/abs/2306.06506

I also have made a Medium blog post with a more summarized review of the methods: https://mazzine.medium.com/measuring-counterfactual-explanations-feature-importance-with-counterplots-41359bc85aea

We hope this new package can be useful to you and we are open for your comments and suggestions, below I made a more technical description of the changes made in this PR.

This commit adds the CounterPlots package to the DiCE CF generation package. Basically, the CounterfactualExplanations object was changed to now include a new method generate_counterplots which takes factual, counterfactual, and DataFrame info to generate the counterfactual analysis. Also, this object now asks for a prediction function, therefore, the explainer_base, dice_tensorflow1, and dice_tensorflow2 now include this function to create the object.

Best wishes,

[rmazzine]

dice_counterplots_example_notebook.zip

The zip above has a Jupyter Notebook that shows how counterplots work with DiCE.

[rmazzine]

Very soon I will make the necessary modifications in the unitary tests.

[codecov-commenter]

Codecov Report

Patch coverage: 24.24% and project coverage change: -1.51 ⚠️

Comparison is base (b1b21ae) 68.33% compared to head (bbf511d) 66.83%.

❗ Your organization is not using the GitHub App Integration. As a result you may experience degraded service beginning May 15th. Please install the Github App Integration for your organization. Read more.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #377      +/-   ##
==========================================
- Coverage   68.33%   66.83%   -1.51%     
==========================================
  Files          27       27              
  Lines        3610     3639      +29     
==========================================
- Hits         2467     2432      -35     
- Misses       1143     1207      +64     

Flag	Coverage Δ
unittests	66.83% <24.24%> (-1.51%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
dice_ml/explainer_interfaces/dice_tensorflow1.py	0.00% <0.00%> (ø)
dice_ml/counterfactual_explanations.py	65.42% <20.00%> (-30.81%)	⬇️
dice_ml/explainer_interfaces/dice_tensorflow2.py	85.06% <100.00%> (ø)
dice_ml/explainer_interfaces/explainer_base.py	88.11% <100.00%> (-0.26%)	⬇️

... and 1 file with indirect coverage changes

☔ View full report in Codecov by Sentry.
📢 Do you have feedback about the report comment? Let us know in this issue.

[amit-sharma]

@rmazzine Thanks for starting this PR. Will you be able to look at the unit tests and make sure that they pass?

Apologies that I missed responding to this earlier.

[rmazzine]

I will close this PR because I've made a better (and easier) solution in PR #402