Home // SIMUL 2025, The Seventeenth International Conference on Advances in System Modeling and Simulation // View article


Traditional Statistics and Machine Learning in Social Network Analysis: A Comparative Reanalysis of Social Network Data on Energy Transition Decisions

Authors:
Mart Verhoog

Keywords: Social Network Analysis; Machine Learning; Traditional Statistics; Comparative Methods; Interpretability and Prediction

Abstract:
The goal of this idea contribution is to provide a systematic head-to-head comparison of regression-based inference and Machine Learning (ML) prediction in applied Social Network Analysis (SNA) for energy transition research, addressing a gap that has not yet been explored. The problem is relevant because methodological choices affect how actor influence and decision-making are interpreted in networked household energy-efficient refurbishments. While regression models offer explanatory clarity, ML models often deliver higher predictive accuracy; yet their joint evaluation in this domain remains missing. This study proposes a structured pipeline combining regression baselines with ML models, such as Random Forests, Support Vector Machines (SVMs), and Gradient Boosting. Model performance will be evaluated using R² and the Receiver Operating Characteristic – Area Under the Curve (ROC-AUC), while interpretability will be assessed through SHapley Additive exPlanations (SHAP) values. The expected outcome is a sharper understanding of trade-offs and complementarities between inference and prediction in energy transition networks, informing methodological integration in computational social science.

Pages: 95 to 96

Copyright: Copyright (c) IARIA, 2025

Publication date: September 28, 2025

Published in: conference

ISSN: 2308-4537

ISBN: 978-1-68558-300-2

Location: Lisbon, Portugal

Dates: from September 28, 2025 to October 2, 2025