Analyzing Household Expenditures with Generalized Random Forests

This study investigates the performance of Generalized Random Forest (GRF), which has been known to be useful in understanding heterogeneous treatment effects (HTE) and non-linear relationships in high-dimensional data. In this paper the performance of GRF was compared with Random Forest (RF), Gener...

Full description

Saved in:
Bibliographic Details
Main Authors: Eriski Isnanda, Khairil Anwar Notodiputro, Kusman Sadik
Format: Article
Language:English
Published: Mathematics Department UIN Maulana Malik Ibrahim Malang 2025-03-01
Series:Cauchy: Jurnal Matematika Murni dan Aplikasi
Subjects:
Online Access:https://ejournal.uin-malang.ac.id/index.php/Math/article/view/30104
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:This study investigates the performance of Generalized Random Forest (GRF), which has been known to be useful in understanding heterogeneous treatment effects (HTE) and non-linear relationships in high-dimensional data. In this paper the performance of GRF was compared with Random Forest (RF), Generalized Linear Mixed Model (GLMM) as continuation of previous study conducted by Athey (2019). The data utilized in this study is from the National Socioeconomic Survey (SUSENAS) to predict household per capita expenditure in West Java, Indonesia. The models are evaluated based on their ability to handle outliers using Winsorization. The results show that RF performed the best, yielding the smallest MSE values, followed by GRF with reasonably good performance, and GLMM with the highest MSE, indicating its limitations in handling non-linear data patterns. These findings indicate that RF is the most accurate method for modeling per capita expenditure in West Java, with recommendations for further research to develop hybrid methods or use more specific random effects in GLMM
ISSN:2086-0382
2477-3344