Title: Low-carbon tourism carbon emission prediction based on LightGBM and model interpreter integration

Authors: Na Zheng

Addresses: School of Automotive Business, Hubei University of Automotive Technology, Shiyan, 442002, China

Abstract: As climate change concerns grow, the tourism industry faces serious carbon emission challenges, especially with urbanisation and rising consumption. Emissions from transportation, accommodation, and cultural activities significantly impact the environment. This study proposes a hybrid model combining light gradient boosting machine (LightGBM) and local interpretable model-agnostic explanations (LIME), optimised using improved particle swarm optimisation (IPSO) and the STIRPAT model. The approach integrates multivariable data to enhance prediction and environmental impact analysis. Experimental results show the model achieves a prediction accuracy of 0.98 and reduces RMSE to 0.13, outperforming traditional XGBoost (RMSE 0.38). Ridge regression coefficients (0.55-0.15) explain the influence of factors on emissions. This hybrid model improves prediction performance and supports carbon management and energy efficiency in tourism. Accurate forecasts enable industries to formulate low-carbon strategies, support sustainable growth, and provide a scientific basis for policymakers, promoting the development of green and sustainable tourism.

Keywords: LightGBM; light gradient boosting machine; LIME; local interpretable model-agnostic explanations; low-carbon tourism; carbon emission prediction; STIRPAT; stochastic impacts by regression on population; affluence; and technology.

DOI: 10.1504/IJTPM.2026.154799

International Journal of Technology, Policy and Management, 2026 Vol.26 No.2, pp.158 - 178

Received: 03 Apr 2025
Accepted: 05 Aug 2025
Published online: 14 Jul 2026 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article