Combination four different ensemble algorithms with the generalized linear model (GLM) for predicting forest fire susceptibility

In this study, the generalized linear model (GLM) and four ensemble methods (partial least squares (PLS), boosting, bagging, Bayesian) were applied to predict forest fire hazard in Chalus Rood watershed Mazandaran Province, Iran. Data from 108 historical events collected through field surveys as basis of analysis. About 70% data used for training models, while remaining 30% was testing. A total 14 environmental, climatic, vegetation variables input features models probability. After conducting a multicollinearity test on independent variables, GLM modeling. The efficiency evaluated using receiver operating characteristic (ROC) curve parameters. Results validation process, based area under ROC (AUC), showed that GLM, PLS-GLM, boosted-GLM, Bagging-GLM, Bayesian-GLM had efficiencies 0.79, 0.75, 0.81, 0.84, 0.85, respectively. results indicated all methods, except PLS algorithm, improved performance modeling hazards watershed, with Bayesian algorithm being most efficient method among them.