Optimal hyperparameter tuning of random forests for estimating causal treatment effects

Recent studies have expanded the focus of machine learning methods like random forests beyond prediction. They have found utility in the area of causal inference by using it to estimate propensity scores. It has also been established in the literature that tuning the hyperparameter values of rando...

Descripción completa

Guardado en:

Detalles Bibliográficos
Autores principales:	Lateef Amusa, Delia North, Temesgen Zewotir
Formato:	article
Lenguaje:	EN
Publicado:	Prince of Songkla University 2021
Materias:	random forest simulation observational studies propensity scores treatment effect causal inference Technology T Technology (General) T1-995 Science Q Science (General) Q1-390
Acceso en línea:	https://doaj.org/article/66666c37aa284b68b145f52e104a4618
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

id	oai:doaj.org-article:66666c37aa284b68b145f52e104a4618
record_format	dspace
spelling	oai:doaj.org-article:66666c37aa284b68b145f52e104a46182021-11-07T16:23:44ZOptimal hyperparameter tuning of random forests for estimating causal treatment effects10.14456/sjst-psu.2021.1320125-3395https://doaj.org/article/66666c37aa284b68b145f52e104a46182021-08-01T00:00:00Zhttps://rdo.psu.ac.th/sjstweb/journal/43-4/12.pdfhttps://doaj.org/toc/0125-3395Recent studies have expanded the focus of machine learning methods like random forests beyond prediction. They have found utility in the area of causal inference by using it to estimate propensity scores. It has also been established in the literature that tuning the hyperparameter values of random forests can improve the estimates of causal treatment effects. We thus address the issue of getting the best out of random forest models by proposing to tune the random forest hyperparameters while maximizing covariate balance. We consider variants of tuning based on a model fit criterion and compare with tuning to chase covariate balance. In a simulation study and empirical application in two case studies, we studied the performance of different tuning implementations, relative to the random forest with default hyperparameters. We find that tuning to chase balance rather than model fit when estimating propensity scores induced better balance in the covariates and produced more accurate treatment effect estimates.Lateef AmusaDelia NorthTemesgen ZewotirPrince of Songkla Universityarticlerandom forestsimulationobservational studiespropensity scorestreatment effectcausal inferenceTechnologyTTechnology (General)T1-995ScienceQScience (General)Q1-390ENSongklanakarin Journal of Science and Technology (SJST), Vol 43, Iss 4, Pp 1004-1009 (2021)
institution	DOAJ
collection	DOAJ
language	EN
topic	random forest simulation observational studies propensity scores treatment effect causal inference Technology T Technology (General) T1-995 Science Q Science (General) Q1-390
spellingShingle	random forest simulation observational studies propensity scores treatment effect causal inference Technology T Technology (General) T1-995 Science Q Science (General) Q1-390 Lateef Amusa Delia North Temesgen Zewotir Optimal hyperparameter tuning of random forests for estimating causal treatment effects
description	Recent studies have expanded the focus of machine learning methods like random forests beyond prediction. They have found utility in the area of causal inference by using it to estimate propensity scores. It has also been established in the literature that tuning the hyperparameter values of random forests can improve the estimates of causal treatment effects. We thus address the issue of getting the best out of random forest models by proposing to tune the random forest hyperparameters while maximizing covariate balance. We consider variants of tuning based on a model fit criterion and compare with tuning to chase covariate balance. In a simulation study and empirical application in two case studies, we studied the performance of different tuning implementations, relative to the random forest with default hyperparameters. We find that tuning to chase balance rather than model fit when estimating propensity scores induced better balance in the covariates and produced more accurate treatment effect estimates.
format	article
author	Lateef Amusa Delia North Temesgen Zewotir
author_facet	Lateef Amusa Delia North Temesgen Zewotir
author_sort	Lateef Amusa
title	Optimal hyperparameter tuning of random forests for estimating causal treatment effects
title_short	Optimal hyperparameter tuning of random forests for estimating causal treatment effects
title_full	Optimal hyperparameter tuning of random forests for estimating causal treatment effects
title_fullStr	Optimal hyperparameter tuning of random forests for estimating causal treatment effects
title_full_unstemmed	Optimal hyperparameter tuning of random forests for estimating causal treatment effects
title_sort	optimal hyperparameter tuning of random forests for estimating causal treatment effects
publisher	Prince of Songkla University
publishDate	2021
url	https://doaj.org/article/66666c37aa284b68b145f52e104a4618
work_keys_str_mv	AT lateefamusa optimalhyperparametertuningofrandomforestsforestimatingcausaltreatmenteffects AT delianorth optimalhyperparametertuningofrandomforestsforestimatingcausaltreatmenteffects AT temesgenzewotir optimalhyperparametertuningofrandomforestsforestimatingcausaltreatmenteffects
_version_	1718443287392550912

Optimal hyperparameter tuning of random forests for estimating causal treatment effects

Ejemplares similares