References & Citations
Mathematics > Optimization and Control
Title: About evaluation metrics for contextual uplift modeling
(Submitted on 1 Jul 2021 (v1), last revised 2 Aug 2021 (this version, v3))
Abstract: In this tech report we discuss the evaluation problem of contextual uplift modeling from the causal inference point of view. More particularly, we instantiate the individual treatment effect (ITE) estimation, and its evaluation counterpart. First, we unify two well studied fields: the statistical ITE approach and its observational counterpart based on uplift study. Then we exhibit the problem of evaluation, based on previous work about ITE and uplift modeling. Moreover, we derive a new estimator for the uplift curve, built on logged bandit feedback dataset, that reduces its variance. We prove that AUUC fails on non randomized control trial (RCT) datasets and discuss some corrections and guidelines that should be kept in mind while using AUUC (re-balancing the population, local importance sampling using propensity score).
Submission history
From: Matthieu Martin [view email][v1] Thu, 1 Jul 2021 15:28:59 GMT (3185kb,D)
[v2] Fri, 2 Jul 2021 08:24:55 GMT (3185kb,D)
[v3] Mon, 2 Aug 2021 09:59:39 GMT (3421kb,D)
Link back to: arXiv, form interface, contact.