Hybrid conditional autoregressive generalized random forest for overdispersed spatial counts: Evidence from Kalimantan wildfire hotspots

Abstract

This paper proposes a negative binomial conditional autoregressive generalized random forest model for overdispersed spatial count data. The framework combines a flexible nonparametric mean function estimated via generalized random forests with conditional autoregressive spatial random effects under a negative binomial likelihood. We evaluate the proposed method through a simulation study and an empirical application to satellitederived wildfire hotspot counts in Kalimantan, Indonesia (September 2024), using rainfall, air temperature, relative humidity, wind speed, and the number of rainless days as covariates. Across simulation scenarios, negative binomial conditional autoregressive generalized random forest yields lower prediction errors than the benchmark negative binomial conditional autoregressive random forest. In the Kalimantan application, the model achieves a root mean squared error of 3.64 and a mean absolute error of 2.11 hotspots per grid cell. Variable importance analysis indicates that air temperature is the most influential predictor of the spatial distribution of hotspots.

Keywords

• Conditional autoregressive
• generalized random forest
• overdispersed count data
• negative binomial
• spatial modelling

Supporting Institution

This study was financially supported by the Indonesian Education Scholarship [BPI ID: 202209090500], Center for Higher Education Funding and Assessment, Ministry of Higher Education, Science, and Technology of the Republic of Indonesia, and Endowment Fund for Education Agency, Ministry of Finance of the Republic of Indonesia.

References

1. [1] H.A. Abood, S. Khatan Ismail and A.F.T. Sabah, Re-weighted least squares: the best negative binomial regression methods in determining the congenital anomalies’ risk factors, Syst. Rev. Pharm. 11 (12), 1956-1963, 2020.
2. [2] G. Alomair, Predictive performance of count regression models versus machine learning techniques: a comparative analysis using an automobile insurance claims frequency dataset, PLoS One 19 (12), 1-12, 2024.
3. [3] J. Argota Sánchez-Vaquerizo and D. Helbing, Spatial econometrics to estimate traffic reduction by transforming office space into housing and other land uses: the case for Barcelona, Cities 162, 1-21, 2025.
4. [4] S. Athey, J. Tibshirani and S. Wager, Generalized random forests, Ann. Stat. 47 (2), 1148-1178, 2019.
5. [5] I. Azis, A. Djuraidah, M. Aidi and A. Sopaheluwakan, Spatial prediction of the number of wildfire hotspots using negative binomial conditional autoregressive random forest, Glob. J. Environ. Sci. Manag. 11 (4), 1-18, 2025.
6. [6] L. Breiman, Random forests, Mach. Learn. 45, 5-32, 2001.
7. [7] M. Chen, J. Liu, B. Chu, D. Zhao, R. Li, T. Chen, Q. Ma, Y. Wang, P. Zhang, H. Li and H. He, Research on the influencing factors of PM2.5 in China at different spatial scales based on machine learning algorithm, Environ. Int. 200, 1-10, 2025.
8. [8] K.A. Coskuner, Assessing the performance of MODIS and VIIRS active fire products in the monitoring of wildfires: a case study in Turkey, iForest 15 (2), 85-94, 2022.

9. [9] İ. Daşdemir, F. Aydın and M. Ertuğrul, Factors affecting the behavior of large forest fires in Turkey, Environ. Manage. 67 (1), 162-175, 2021.
10. [10] D.J. Fitri, A. Djuraidah and H. Wijayanto, Bayesian conditional negative binomial autoregressive model: a case study of stunting on Java Island in 2021, Commun. Math. Biol. Neurosci. 2024, 1-17, 2024.
11. [11] J.K. Frank, T. Suesse and A. Brenning, An assessment of spatial random forests for environmental mapping: the case of groundwater nitrate concentration, Environ. Model. Softw. 193, 1-14, 2025.
12. [12] L. Giglio, J. Descloitres, C.O. Justice and Y.J. Kaufman, An enhanced contextual fire detection algorithm for MODIS, Remote Sens. Environ. 87 (2), 273-282, 2003.
13. [13] H. Hersbach, B. Bell, P. Berrisford, S. Hirahara, A. Horányi, J. Muñoz-Sabater, J. Nicolas, C. Peubey, R. Radu, D. Schepers, A. Simmons, C. Soci, S. Abdalla, X. Abellan, G. Balsamo, P. Bechtold, G. Biavati, J. Bidlot, M. Bonavita and J.-N. Thépaut, The ERA5 global reanalysis, Q. J. R. Meteorol. Soc. 146 (730), 1999-2049, 2020.
14. [14] M.W. Jones, J.T. Abatzoglou, S. Veraverbeke, N. Andela, G. Lasslop, M. Forkel, A.J.P. Smith, C. Burton, R.A. Betts, G.R. van der Werf, S. Sitch, J.G. Canadell, C. Santín, C. Kolden, S.H. Doerr and C. Le Quéré, Global and regional trends and drivers of fire under climate change, Rev. Geophys. 60 (3), 1-76, 2022
15. [15] W.S. Karurung, K. Lee and W. Lee, Assessment of forest fire vulnerability prediction in Indonesia: seasonal variability analysis using machine learning techniques, Int. J. Appl. Earth Obs. Geoinf. 138, 1-16, 2025.
16. [16] J. Kim, A.B. Lawson, B. Neelon, J.E. Korte, J.M. Eberth and G. Chowell, Evaluation of Bayesian spatiotemporal infectious disease models for prospective surveillance analysis, BMC Med. Res. Methodol. 23 (1), 1-13, 2023.
17. [17] A. Kmoch, C.T. Harrison, J. Choi and E. Uuemaa, Spatial autocorrelation in machine learning for modelling soil organic carbon, Ecol. Inform. 86, 1-21, 2025.
18. [18] F. Lu, G. Zhang, T.Wang, Y. Ye and Q. Zhao, Geographically weighted random forest based on spatial factor optimization for the assessment of landslide susceptibility, Remote Sens. 17 (9), 1-28, 2025.
19. [19] L. Luo, D. Apps, S. Arcand, H. Xu, M. Pan and M. Hoerling, Contribution of temperature and precipitation anomalies to the California drought during 2012–2015, Geophys. Res. Lett. 44 (7), 3184-3192, 2017.
20. [20] C. MacBride, V. Davies and D. Lee, A spatial autoregressive random forest algorithm for small-area spatial prediction, Ann. Appl. Stat. 19 (1), 485-504, 2025.
21. [21] Y.C. MacNab, Adaptive Gaussian Markov random field spatiotemporal models for infectious disease mapping and forecasting, Spat. Stat. 53, 1-23, 2023.
22. [22] T.W. Mas’oed, S. Nurdiati, A. Sopaheluwakan, M.K. Najib and A. Salsabila, Modeling fire hotspots in Kalimantan, Indonesia using nested 3-copula regression based on precipitation and dry days during different ENSO phases, Geographia Technica 19 (2), 264-281, 2024.
23. [23] P. McCullagh and J.A. Nelder, Generalized Linear Models, 2nd ed., Chapman & Hall/CRC, 1989.
24. [24] C.A. Mushagalusa, A.B. Fandohan and R. Glèlè Kakaï, Random forest and spatial cross-validation performance in predicting species abundance distributions, Environ. Syst. Res. 13 (1), 1-24, 2024.
25. [25] F. Mutiso, J.L. Pearce, S.E. Benjamin-Neelon, N.T. Mueller, H. Li and B. Neelon, Bayesian negative binomial regression with spatially varying dispersion: modelling COVID-19 incidence in Georgia, Spat. Stat. 52, 1-19, 2022.
26. [26] M. Pascolini-Campbell, J.B. Fisher, K. Cawse-Nicholson, C.M. Lee and N. Stavros, Assessment of spatial autocorrelation and scalability in fine-scale wildfire random forest prediction models, Sci. Rep. 15 (1), 1-15, 2025.
27. [27] Z. Qin, Q. Peng, C. Jin, J. Xu, S. Xing, P. Zhu and Y. Yang, Geographically weighted random forest fusing multi-source environmental covariates for spatial prediction of soil heavy metals, Environ. Pollut. 385, 2025.
28. [28] A. Schmidt, L.M. Ellsworth, G.A. Boisen, N. Novita, A. Malik, A. Gangga, I. Albar, A.D. Nurhayati, R.P. Ritonga, A. Asyhari and J.B. Kauffman, Fire frequency, intensity, and burn severity in Kalimantan’s threatened peatland areas over two decades, Front. For. Glob. Change 7, 1-16, 2024.
29. [29] J. Stoklosa, R.V. Blakey and F.K.C. Hui, An overview of modern applications of negative binomial modelling in ecology and biodiversity, Diversity 14 (5), 1-25, 2022.
30. [30] S. Ullah, M.A.K. Barakzai and T. Xie, Performance of a negative binomial-GLM in spatial scan statistic: a case study of low-birth weights in Pakistan, Geospat. Health 19 (2), 2024.
31. [31] A. Usup and H. Hayasaka, Peatland fire weather conditions in central Kalimantan, Indonesia, Fire 6 (5), 1-12, 2023.
32. [32] Y. Wang, X. Chen and F. Xue, A review of Bayesian spatiotemporal models in spatial epidemiology, ISPRS Int. J. Geo-Inf. 13 (3), 1-29, 2024.
33. [33] Z. Xu, D. Liu and L. Yan, Temperature-based fire frequency analysis using machine learning: A case of Changsha, China, Clim. Risk Manag. 31, 100276, 2021.