Type I Tobit Gaussian process

Abstract

We present a Gaussian process (GP) approach to regression that can handle data subject to censoring. Since the model is not analytically tractable we use Expectation propagation to perform approximate inference on it.

Code

The Matlab code can be downloaded here. It has been tested with the free software GPstuff v4.5. It can also be used with the Laplace approximation.

References

Perry Groot, Peter Lucas. Gaussian process regression with censored data using Expectation propagation, PGM 2012. [PDF] [Supplement]