Optimizing feature selection and remote sensing classification with an enhanced machine learning method

The Journal of Supercomputing, Год журнала: 2025, Номер 81(2)

Опубликована: Янв. 6, 2025

Язык: Английский

---

Assessment of landslide occurrence and prediction of susceptible zone based on GIS along national highway 37, Manipur, India

Sadhana, Год журнала: 2024, Номер 49(1)

Опубликована: Фев. 15, 2024

Язык: Английский

---

Development of risk maps for flood, landslide, and soil erosion using machine learning model

Natural Hazards, Год журнала: 2024, Номер unknown

Опубликована: Июнь 9, 2024

Язык: Английский

---

Urban Heat Island Mitigation in Tehran: District-based Mapping and Analysis of Key Drivers

Sustainable Cities and Society, Год журнала: 2025, Номер unknown, С. 106338 - 106338

Опубликована: Март 1, 2025

Язык: Английский

---

Utilising Machine Learning Approaches for Enhanced Landslide Susceptibility Mapping in Sikkim, India

Geological Journal, Год журнала: 2025, Номер unknown

Опубликована: Апрель 10, 2025

ABSTRACT Landslides pose significant hazards in the mountainous region of Sikkim, India, necessitating accurate susceptibility mapping to mitigate risks. This study applies four machine learning models: Boosted Tree (BT), Gradient Boosting Machine (GBM), K‐Nearest Neighbour (KNN), and Multilayer Perceptron (MLP) develop a detailed landslide map. Feature selection was performed using correlation analysis, Boruta model, multicollinearity tests, which identified 13 key conditioning factors based on 1456 inventory points. The GBM model demonstrated highest predictive performance with an AUC 0.99, followed by BT (AUC: 0.965), MLP 0.940), KNN 0.895) testing dataset. confusion matrix validation confirmed that outperformed other models, achieving F1 score (0.894) accuracy (89.4%), 0.874 87.8%. displayed lower performance, showing 0.724 72.6%, significantly underperforming 0.096 48.6%. Statistical significance Wilcoxon Signed‐Rank Test revealed differences between ( p = 0.018), while pairs exhibited no statistically differences. Additionally, variable importance analysis highlighted Diurnal Temperature Range (DTR) as most critical factor influencing occurrence (43.99%), elevation (21.59%). These findings provide valuable insights for policymakers government authorities, enabling them take necessary measures effective management vulnerable areas confirming efficacy models geohazard assessments.

Язык: Английский

---

Assessing Land Use and Land Cover Changes in Karnataka’s Western Ghats Using the GeoML-LULC Framework

Journal of The Institution of Engineers (India) Series A, Год журнала: 2025, Номер unknown

Опубликована: Май 26, 2025

Язык: Английский

---

Developing a hybrid deep learning model with explainable artificial intelligence (XAI) for enhanced landslide susceptibility modeling and management

Natural Hazards, Год журнала: 2023, Номер 120(4), С. 3719 - 3747

Опубликована: Дек. 18, 2023

Язык: Английский

---

Assessing the impact of RCP4.5 and RCP8.5 scenarios on landslide susceptibility mapping using support vector machine: A case study of Penang Island, Malaysia

Physics and Chemistry of the Earth Parts A/B/C, Год журнала: 2023, Номер 133, С. 103496 - 103496

Опубликована: Ноя. 14, 2023

Язык: Английский

---

Tourism Suitability Assessment in Malbazar Block using principal component analysis and analytical hierarchy process

Environment Development and Sustainability, Год журнала: 2024, Номер unknown

Опубликована: Дек. 26, 2024

Язык: Английский

---

Learning Ground Displacement Signals Directly from InSAR-Wrapped Interferograms

Sensors, Год журнала: 2024, Номер 24(8), С. 2637 - 2637

Опубликована: Апрель 20, 2024

Monitoring ground displacements identifies potential geohazard risks early before they cause critical damage. Interferometric synthetic aperture radar (InSAR) is one of the techniques that can monitor these with sub-millimeter accuracy. However, using InSAR technique challenging due to need for high expertise, large data volumes, and other complexities. Accordingly, development an automated system indicate directly from wrapped interferograms coherence maps could be highly advantageous. Here, we compare different machine learning algorithms evaluate feasibility achieving this objective. The inputs implemented models were pixels selected filtered-wrapped Sentinel-1, a threshold. outputs same labeled as fast positive, negative, undefined movements. These labels assigned based on velocity values measurement points located within pixels. We used Parallel Small Baseline Subset service European Space Agency's GeoHazards Exploitation Platform create necessary interferograms, coherence, deformation maps. Subsequently, applied high-pass filter separate displacement signal atmospheric errors. successfully identified patterns associated slow movements by discerning unique distributions matrices representing each movement class. experiments included three case studies (from Italy, Portugal, United States), noted their sensitivity landslides. found Cosine K-nearest neighbor model achieved best test It important note sets not merely hidden parts training set region but also adjacent areas. further improved performance pseudo-labeling, approach aimed at evaluating generalizability robustness trained beyond its immediate environment. lowest accuracy algorithm was 80.1%. Furthermore, ArcGIS Pro 3.3 truth predictions visualize results better. comparison explore indications affecting main roads in studied area.

Язык: Английский

---