Using Representation Balancing to Learn Conditional-Average Dose Responses from Clustered Data

Abstract

Estimating the response to an intervention with an associated dose conditional on a unit's covariates, the "conditional-average dose response" (CADR), is a relevant task in a variety of domains, from healthcare to business, economics, and beyond. Estimating such a response is challenging for several reasons: Firstly, it typically needs to be estimated from observational data, which can be confounded and negatively affect the performance of intervention response estimators used for counterfactual inference. Secondly, the continuity of the dose prevents the adoption of approaches used to estimate responses to binary-valued interventions. That is why the machine learning (ML) community has proposed several tailored CADR estimators. Yet, the proposal of most of these methods requires strong assumptions on the distribution of data and the assignment of interventions, which go beyond the standard assumptions in causal inference. Whereas previous works have so far focused on smooth shifts in covariate distributions across doses, in this work, we will study estimating CADR from clustered data and where different doses are assigned to different segments of a population. On a novel benchmarking dataset, we show the impacts of clustered data on model performance. Additionally, we propose an estimator, CBRNet, that enables the application of representation balancing for CADR estimation through clustering the covariate space and a novel loss function. CBRNet learns cluster-agnostic and hence dose-agnostic covariate representations through representation balancing for unbiased CADR inference. We run extensive experiments to illustrate the workings of our method and compare it with the state of the art in ML for CADR estimation.