Spatial Analysis of Flood Hazard Zoning Map Using Novel Hybrid Machine Learning Technique in Assam, India

Twenty-two flood-causative factors were nominated based on morphometric, hydrological, soil permeability, terrain distribution, and anthropogenic inferences further analyzed through the novel hybrid machine learning approach of random forest, support vector machine, gradient boosting, naïve Bayes, decision tree (ML) models. A total 400 flood nonflood locations acted as target variables hazard zoning map. All operative in this study tested using variance inflation factor (VIF) values (<5.0) Boruta feature ranking (<10 ranks) for FHZ maps. The model along with RF GBM had sound maps area. area under receiver operating characteristics (AUROC) curve statistical matrices such accuracy, precision, recall, F1 score, gain lift applied to assess performance. 70%:30% sample ratio training validation standalone models concerning AUROC value showed results all ML models, (97%), SVM (91%), NB (96%), DT (88%), (97%). also suitability RF, GBM, developing