Testing for Causal Effects based on Event Processes

20.03.2026 11:15 – 12:15

RESEARCH INSTITUTE FOR STATISTICS AND INFORMATION SCIENCE SEMINARS

ABSTRACT

In the context of disease progression analysis, testing for possible causal effects of a time-dependent treatment assigned to a given population relies on the ability to construct and compute a conditional local independence test. This is relevant in many applications that analyze high-dimensional and time-dependent observational data to understand causal relations, for example in econometrics, social sciences and epidemiology.
In this work, we propose a nonparametric framework for testing possible causal effects by introducing SHAPE: a novel usage of lasso regularization for computing Sparse and High-dimensional Adaptive Projections of Events. We prove new finite-sample guarantees for the estimation and prediction errors under suitable regularity conditions. Although our results are theoretical, we demonstrate their practical relevance for designing efficient algorithmic procedures. The results will further be applied to learn causal graphs in this setting.
(Jointly with Prof. Niels R. Hansen (Copenhagen University, Denmark))

Lieu

Bâtiment: Uni Mail

Boulevard du Pont-d'Arve 40
1205 Geneva

Room M 4220, 4th floor

entrée libre