A Python implementation for estimating the predictive relevance of input variables for Gaussian process models. The repository contains functions that use methods from the paper Paananen et al. (2019). There is also a small demo that demonstrates the toy example from the paper. The functions utilize the GPy framework (https://github.com/SheffieldML/GPy) for Gaussian process inference.
Paananen, T., Piironen, J., Andersen, M.R., and Vehtari, A. (2019). Variable selection for Gaussian processes via sensitivity analysis of the posterior predictive distribution. In Proceedings of the 22nd International Conference on Artificial Intelligence and Statistics (AISTATS), PMLR 89: 1743-1752. (Online)