alibi
alibi copied to clipboard
Algorithms for explaining machine learning models
## Issue The `save` method of the `PartialDependenceVariance` explainer returns the following error: ``` AttributeError: 'PartialDependenceVariance' object has no attribute 'predictor' ``` This is because the `predictor` attribute belongs to...
There seems to be an edge case which is not considered in our implementation. For `ligthgbm` only the path-dependent method is supported when categorical features exists (i.e. in a `pd.DataFrame`...
When saving explainers, the `predictor` attribute is set to `None` to avoid attempting to save it, and it is then given to `load_explainer` at load time: https://github.com/SeldonIO/alibi/blob/c2da0124983b2a2b3d52920428c45528eb180d2e/alibi/saving.py#L110-L112 However, some explainers,...
I have the following error when trying to call explain on an AnchorTabular fitted on a mixed of numerical and categorical data. On alibi 0.9.2. ``` Error Traceback (most recent...
Problem Statement: 1. I am trying to fit a dataset with categorical variables, similar to adult dataset example. 2. However I want to fit the explainer model for multiple datasets...
Error message is `NotImplementedError: numpy() is only available when eager execution is enabled.` so seems `tensorflow` related. Started failing (with `tensorflow 2.11.1`): https://github.com/SeldonIO/alibi/actions/runs/4647731609/jobs/8224890365 Passing run 3 weeks ago (with `tensorflow...
I have followed this example : https://docs.seldon.io/projects/alibi/en/stable/examples/cfrl_adult.html I am generating counterfactual for two class 0 & 1. But for my example in cell 18 : cf_pd ,which is generated counterfactual...
The are multiple ways to deal with one hot encoding for the categorical features when using the `KernelShap` explainer. * The first method is straightforward and incorporates the preprocessor into...
The `sparse` argument in the `OneHotEncoder` is deprecated and will be fully replaced by `spares_output`. In 1.4., `sparse` output will be removed which will require an update to this [notebook](https://github.com/SeldonIO/alibi/blob/master/doc/source/examples/cfproto_cat_adult_ohe.ipynb).
`sklearn 1.2` included an new parameter, `normalized_stress`, to the [`manifold.MDS`](https://scikit-learn.org/stable/modules/generated/sklearn.manifold.MDS.html#sklearn.manifold.MDS). Until 1.4., the default value is `warn` which will print a warning, but then replaced by `False`. Following 1.4. the...