Maximilian
Maximilian
The examples currently sitting in the root of the projects should be moved into notebooks and run to showcase the explanations. For this the visualization (#29) and the validators (#24)...
Add specific unit tests for the validators that show that they are working as intended. (Torch model, sklearn model, river models, custom loss functions, river metrics)
The storage module needs to be documented further. At least all public classes and public class methods need to be well documented.
The imputer module needs to be documented further. At least all public classes and public class methods need to be well documented.
Extend Capabilities of the visualization submodule that has sensible defaults for iPFI and iSAGE (e.g. performance of SAGE in terms of marginal and model loss.)
The Bar Plot presented [here](https://shapiq.readthedocs.io/en/latest/notebooks/treeshapiq_lightgbm.html#global-feature-(interaction)-importance) does not work properly when one uses the bike sharing data set as in the reworked notebook [here](https://github.com/mmschlk/shapiq/blob/188-rework-and-reprganize-existing-notebook-tutorials/docs/source/notebooks/treeshapiq_lightgbm.ipynb).
Currently, the stacked bar plot is only usable with these `n_shapley_values_pos: dict` and `n_shapley_values_neg: dict` parameters. This is by now outdated and shoud be replaced with an `InteractionValues` object and...
Add a notebook focusing on the computation of Shapley values. This notebook should show that not only interactions are part of shapiq.