Solving Multi-Level Treatments causal discovery problem with multinomial logistic regression and causal forest

Conference: ICMLCA 2021 - 2nd International Conference on Machine Learning and Computer Application
12/17/2021 - 12/19/2021 at Shenyang, China

Proceedings: ICMLCA 2021

Pages: 4Language: englishTyp: PDF

Personal VDE Members are entitled to a 10% discount on this title

Authors:
Sun, Leting (School of Statistics and Information, Shanghai University of International Business and Economics, Shanghai,China)

Abstract:
In this article, wecombine the multinomial logistic regression and causal forest to estimate average treatment effects in observational studies, in situations where there are multi-level treatments. The method of using propensity score to group the covariables into groups and mathcing them to compute the average causal effect have been used widely among in popular papers in analyzing binary treatment problems. However, the methods used in these literatures, for example the causal forest, are based on weak unconfoundedness assumption. As a result, these methods couldn’t be simply applied to multi-level treatments problems. So this work used the idea of strong unconfoundedness in Propensity Score Matching and Subclassification in Observational Studies with Multi-Level Treatments. This work use multinomial logistic regression to do the propensity score subclassification and the causal forest asthe matching meth.