Handling data imbalance in machine learning based landslide susceptibility mapping: a case study of Mandakini River Basin, North-Western Himalayas

Gupta, Sharad; Shukla, Dericks Praise

doi:10.1007/s10346-022-01998-1

Cited by 28 publications

(6 citation statements)

References 76 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A landslide problem is a binary classification problem (landslide or non-landslide), and the confusion matrix and ROC curve are the most commonly used evaluation indexes 44 – 47 . In the confusion matrix, the classification of the different sample categories can be clearly seen.…”

Section: Methodsmentioning

confidence: 99%

Effects of non-landslide sampling strategies on machine learning models in landslide susceptibility mapping

Gu,

Duan,

Wang

et al. 2024

Sci Rep

View full text Add to dashboard Cite

This study aims to explore the effects of different non-landslide sampling strategies on machine learning models in landslide susceptibility mapping. Non-landslide samples are inherently uncertain, and the selection of non-landslide samples may suffer from issues such as noisy or insufficient regional representations, which can affect the accuracy of the results. In this study, a positive-unlabeled (PU) bagging semi-supervised learning method was introduced for non-landslide sample selection. In addition, buffer control sampling (BCS) and K-means (KM) clustering were applied for comparative analysis. Based on landslide data from Qiaojia County, Yunnan Province, China, collected in 2014, three machine learning models, namely, random forest, support vector machine, and CatBoost, were used for landslide susceptibility mapping. The results show that the quality of samples selected using different non-landslide sampling strategies varies significantly. Overall, the quality of non-landslide samples selected using the PU bagging method is superior, and this method performs best when combined with CatBoost for predicting (AUC = 0.897) landslides in very high and high susceptibility zones (82.14%). Additionally, the KM results indicated overfitting, displaying high accuracy for validation but poor statistical outcomes for zoning. The BCS results were the worst.

show abstract

Section: Methodsmentioning

confidence: 99%

Effects of non-landslide sampling strategies on machine learning models in landslide susceptibility mapping

Gu,

Duan,

Wang

et al. 2024

Sci Rep

View full text Add to dashboard Cite

show abstract

“…Previous studies have shown that SVM model has excellent performance of balancing data samples [32], [33]. In this study, SVM is selected for the base classifier to better find the hyperplane between classification samples.…”

Section: A Sample Optimization Balanced Bagging Methodsmentioning

confidence: 99%

Thaw Slump Susceptibility Mapping Based on Sample Optimization and Ensemble Learning Techniques in Qinghai-Tibet Railway Corridor

He,

Huo,

Gao

et al. 2024

IEEE J. Sel. Top. Appl. Earth Observations Remote Sensing

View full text Add to dashboard Cite

Thaw slump susceptibility mapping (TSSM) of Qinghai-Tibet Railway corridor (QTRC) is the prerequisite and basis for disaster assessment and prevention of permafrost projects. The objective of this study is to construct ensemble learning models based on single classifier models to generate the TSSM of the QTRC, compare and verify the performance of the models, and further explore the relationship between the high susceptibility area and environmental factors of the QTRC. The collinearity analysis was carried out by selecting 14 thaw slump conditioning factors (TSCFs). We used the balance bagging method for sample optimization, and the data set was divided into 70% training set and 30% verification set. Convolutional neural network (CNN), multilayer perceptron (MLP), support vector regression (SVR), random forest (RF) single classifiers were selected to construct blending and stacking ensemble learning models for the TSSM. The results showed that there was no collinearity among the 14 TSCFS. The comparison of model performance revealed that all models had good performance, but the constructed stacking and blending ensemble learning models had stable performance and high prediction accuracy for TSSM. The stacking ensemble learning model had the best effect, and the area under curve (AUC) value of receiver operating characteristic (ROC) curve reached 0.9607. It showed that the generated TSSM of QTRC based on stacking ensemble learning model had the highest reliability. The QTRC has local areas with high thaw slump susceptibility, mainly concentrated in the permafrost areas with high altitude, high slope, adjacent faults, sparse vegetation, ice and snow and the more cumulative precipitation.

show abstract

“…Landslides are deadly and unpredictable type of natural disaster, which bring serious damage to the properties and human life [1][2][3][4][5]. Landslides are complex phenomena influenced by numerous criteria such as geological conditions, geomorphology, climate, and anthropology activities [4].…”

Section: Introductionmentioning

confidence: 99%

Integrating Remote Sensing, GIS and Machine Learning Approaches in Evaluation of Landslide Susceptibility in Mountainous Region of Nghe An Province, Vietnam

Thi Tuyen,

Thi An,

Van An

et al. 2024

IOP Conf. Ser.: Earth Environ. Sci.

View full text Add to dashboard Cite

This study applied remote sensing methods combining GIS and machine learning (ML) in landslide assessment and zonation for the western mountainous area of Nghe An province, Vietnam. Factors affecting landslide susceptibility are analyzed and included in the assessment model including terrain elevation, slope, aspect, flow accumulation, geomorphology, profile curvature, Topographic Position Index (TPI), fault density, road density, rainfall and land use. A field survey was conducted on July, 2023 to collect the ground truth data of landslide areas in Nghe An and used as input for the training and validating process of landslide model with ratios of 70 and 30 percentage. The landslide estimation algorithms which derived from the machine learning approach including Support Vector Machine, Random Forest, and Logistic Regression have been investigated with 11 input layers and field survey training data. The results indicated that among the causative parameters of landslides in the study area, the most important factor was the Standardized Precipitation Index, derived from the rainfall data. Additionally, traffic, terrain slope, and elevation were also significant factors. In terms of the landslide estimation algorithms, the Random Forest model exhibited the highest accuracy for mapping landslide susceptibility in the western mountainous region of Nghe An province, with a correlation coefficient (R2) of 0.97. The research findings demonstrate the effectiveness of integrating remote sensing, GIS, and ML techniques for landslide research in mountainous areas of Vietnam. This approach provides valuable insights on landslide susceptibility, and a better understanding of landslide dynamics in the study area.

show abstract

Handling data imbalance in machine learning based landslide susceptibility mapping: a case study of Mandakini River Basin, North-Western Himalayas

Cited by 28 publications

References 76 publications

Effects of non-landslide sampling strategies on machine learning models in landslide susceptibility mapping

Effects of non-landslide sampling strategies on machine learning models in landslide susceptibility mapping

Thaw Slump Susceptibility Mapping Based on Sample Optimization and Ensemble Learning Techniques in Qinghai-Tibet Railway Corridor

Integrating Remote Sensing, GIS and Machine Learning Approaches in Evaluation of Landslide Susceptibility in Mountainous Region of Nghe An Province, Vietnam

Contact Info

Product

Resources

About