Repository files navigation

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}}

In thesurvshap_package directory, you will find the code forsurvshap Python package, which contains the implementation of the SurvSHAP(t) method.Now you can also easily install it fromPyPI:

pip install survshap

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

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

• survlime.py is theSurvLIME method implementation
• survnam directory contains theSurvNAM method implementation (based onJia-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 directory contains the datasets used in experiments

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

• plots directory contains Figures in.pdf format

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