SAE 2021 - Using Random Forests in Small Area Estimation

Abstract

In the talk, I promote the use of random forests as versatile tools for estimating spatially disaggregated indicators in the presence of small area-specific sample sizes. Small area estimators are predominantly conceptualized within the regression-setting and rely on linear mixed models to account for the hierarchical structure of the survey data. In contrast, machine learning methods offer non-linear and non-parametric alternatives, combining excellent predictive performance and a reduced risk of model-misspecification. Mixed effects random forests combine advantages of regression forests with the ability to model hierarchical dependencies. I discussed a coherent framework based on mixed effects random forests for estimating small area averages and proposed a non-parametric bootstrap estimator for assessing the uncertainty of the estimates. The presentation included the evaluation of our proposed methodology based on a model-based simulation comparing the proposed methodology to traditional regression-based approaches for estimating small area averages.

Dr. Patrick Krennmair

Research Associate in Applied Statistics

I am working as a research associate at the Chair of Applied Statistics at Freie Universität Berlin and as a consultant for the statistical consulting unit fu:stat.