Mateusz Krzyziński
Mateusz Krzyziński
The `metric_function` can't be used as a loss function in hyperparameter optimization - it would be useful to be able to use this function in this process as well.
Related to #65 I found information that Spearman correlation coefficient is used for numerical variables and Cramer's V for categorical ones. It would be nice to be able to set...
Adding [support](https://github.com/ModelOriented/survex/issues/75) for survival models other than `ranger` (see [#24](https://github.com/ModelOriented/treeshap/issues/24)). The source code for the `ranger` can be followed. Functionality for `randomForestSRC` should be relatively easy to implement.
When the render function is called again, the new widget appears on top of the old widget (the old one is not removed). Added two lines fix it. This can...
Potentially, version 1.0 of `survshap` could be published as a separate package on PyPI. **What needs to be done?** - [ ] adding unit tests - [ ] adding missing...
By default, limit the number of timestamps by generating new vector based on times in the background data. **Possible methods of generation:** - equally spaced points (`"uniform"` in `survex`) -...
- [ ] implement `shap.TreeExplainer` for local explanations (`PredictSurvSHAP`) - [ ] implement `shap.TreeExplainer` for global explanations (`ModelSurvSHAP`) - [ ] check the other explainer types from `shap` library