Evaluation of Causal Inference Models to Access Heterogeneous Treatment Effect
Abstract: Causal inference has gained popularity over the last years due to the ability to see through
correlation and find causal relationship between covariates. There are a number of methods
that were created to this end, but there is not a systematic benchmark between those
methods, including the benefits and drawbacks of using each one of them. This research
compares a number of those methods on how well they access the heterogeneous treatment
effect using a variety of synthetically created data sets, divided between low-dimensional
and high-dimensional covariates and increasing complexity between the covariates and the
target. We compare the error between those method and discuss in which setting and
premises each method is better suited.
Submission Length: Regular submission (no more than 12 pages of main content)
Assigned Action Editor: ~Mingming_Gong1
Submission Number: 870
Loading