Explaining Machine Learning Survival Models Event

Time:

11:00 - 12:00

Date:

2024-11-07 00:00:00

Venue:

Ketley Room, Building 54

Book this Event

Event details

In this work, we introduce JointLIME, a novel interpretation method for explaining black-box survival models with endogenous time-varying covariates (TVCs).

JointLIME minimises the distances between survival functions predicted by the black-box survival model and those derived from the joint model. The outputs of this minimisation problem serve as explanations to quantify their impact on survival predictions. JointLIME uniquely incorporates endogenous time varying covariates using a spline-based model. We illustrate the explanation results of JointLIME using a US mortgage dataset and compare them with those of SurvLIME.

Speaker information

Belen Martín-Barragán is Reader in Management Science at The University of Edinburgh. Her research lies at the interface between Data Science and Mathematical Programming, with a special interest on Explainable Artificial Intelligence. In a board sense, her research interests focus on using or developing variants of Operational Research techniques and applying them to data analysis problems, such as classification and clustering. Her work has appeared in a variety of top-ranked journals such as European Journal of Operational Research, Risk Analysis, INFORMS Journal on Computing, Discrete Applied Mathematics, Computers and Operations Research. She is also contributed to other areas such as statistics (Journal of Applied Statistics) and economics (Economic Modelling).