EURO 2024 Copenhagen
Abstract Submission

EURO-Online login

4094. Algorithm-Enhanced Fuel Poverty Prediction Using Household Characteristics, Socio-Economic Factors, and Clustering Analysis

Invited abstract in session TA-14: Enhanced statistical methods for energy challenges, stream Energy Markets.

Tuesday, 8:30-10:00
Room: 16 (building: 116)

Authors (first author is the speaker)

1. Reinhard Madlener
School of Business and Economics / E.ON Energy Research Center, RWTH Aachen University
2. Rahil dejkam
School of Business and Economics / E.ON Energy Research Center, RWTH Aachen University

Abstract

This paper contributes to the literature on predicting fuel poverty risk by applying clustering analysis to identify fuel-poor households. The study develops a novel approach to clustering-based XGBoost model by integrating data from a representative household survey in the UK (N=11'974). The unobserved heterogeneity in fuel poverty households survey hides certain relationships between the contributory features in survey and fuel poverty. This study explores the application of the k-prototypes clustering method to group households into heterogenous clusters. The study segments the entire dataset into three clusters. Four XGBoost models were developed and applied to the entire dataset and each cluster to predict fuel poverty in households. The list of input features includes housing characteristics, socio economic features, and energy-cost variables. Additionally, for analyzing all features the shapley additive explanations (SHAP) method has applied, showing the different degrees of effects of each feature on fuel poverty in the entire dataset. The k-prototypes clustering-based XGBoost model is a promising approach to identifying the households at the risk of fuel poverty. Policymakers can gain new insights by identiying key socio-economic factors to alleviate fuel poverty.

Keywords

Status: accepted

Back to the list of papers