SurvSHAP(t)

This repository contains data and code for the article:

M. Krzyziński, M. Spytek, H. Baniecki, P. Biecek. SurvSHAP(t): Time-dependent explanations of machine learning survival models. Knowledge-Based Systems, 262:110234, 2023. https://doi.org/10.1016/j.knosys.2022.110234

@article{survshap,
    title = {SurvSHAP(t): Time-dependent explanations of machine learning survival models},
    author = {Mateusz Krzyziński and Mikołaj Spytek and Hubert Baniecki and Przemysław Biecek},
    journal = {Knowledge-Based Systems},
    volume = {262},
    pages = {110234},
    year = {2023}
}

Implementations

In the survshap_package directory, you will find the code for survshap Python package, which contains the implementation of the SurvSHAP(t) method. Now you can also easily install it from PyPI:

pip install survshap

NOTE: SurvSHAP(t) and SurvLIME are also implemented in the survex R package, along with many more explanation methods for survival models. survex offers explanations for scikit-survival models loaded into R via the reticulate package.

Additional materials

In addition to the package, the repository also contains the materials used for the article (in the paper directory).

other_codes

• survlime.py is the SurvLIME method implementation
• survnam directory contains the SurvNAM method implementation (based on Jia-Xiang Chengh implementation)
• data_generation.R is the code for synthetic censored data generation (for Experiments 1 and 2)
• plots.R is the code for creating Figures from the article

data

• data directory contains the datasets used in experiments

experiments

• experiments directory contains Jupyter Notebooks (*.ipynb files) with code of the conducted experiments

plots

• plots directory contains Figures in .pdf format

results

• results directory contains results of the conducted experiments stored in .csv files