Model-assisted estimation through random forests in finite population sampling

02/22/2020 ∙ by Mehdi Dagdoug, et al. ∙ 0

Surveys are used to collect data on a subset of a finite population. Most often, the interest lies in estimating finite population parameters such as population totals and means. In some surveys, auxiliary information is available at the population level. This information may be incorporated in the estimation procedures to increase their precision. Model-assisted procedures may be based on parametric or nonparametric models. In this paper, we propose a new class of model-assisted procedures based on random forests based on partitions built at the population level as well as at the sample level. We derive associated variance estimators and we establish the theoretical properties of the proposed procedures. A model-calibration procedure that has the ability to handle multiple survey variables is discussed. Finally, the results of a simulation study suggest that the proposed point and estimation procedures perform well in term of bias, efficiency and coverage in a wide variety of settings.



There are no comments yet.


page 1

page 2

page 3

page 4

This week in AI

Get the week's most popular data science and artificial intelligence research sent straight to your inbox every Saturday.