Predictive Modeling of Landslide Using Machine Learning

Natural disasters, inevitably have a devastating impact on our lifestyle, often disturbing the very fabric of our existence. Among them, landslide is one of the threatening ones due to the seriousness of damage that could result not only in human resource but also in the natural resource supporting our earth around the globe. Therefore, the constantly increasing rate at which people are settling in mountainous areas, whether residential, recreational, or for economic reasons, has led to escalating safety issues that cannot be ignored. Such geospatial information encompasses a rich source of information regarding the Earth's surface and its various features, which allows understanding the complexity of topography, the land cover, and state of infrastructure in different regions. We are able to find far more profound insights into our observations, make very fair predictions, and react well to an extraordinary range of complex problems ranging from medicine and finance to social phenomena and even global climate changes. Many distinct algorithms such as support vector machines, decision trees, artificial neural networks, logistic regression, random forests, and gradient boosting algorithms are applied to predict landslide likelihoods across different geographical regions. Thus, the CNN-2D algorithm has been utilized in existing systems and the maximum accuracy rate achieved is reportedly 82.56%. We achieved an accuracy of 84.25% with our developed MLP model. This altogether outperformed the performance metrics developed by the already existing methods in use.