Research: Machine Learning-Based Prediction of Slope Failure Using Integrated Geotechni...

Research question: How accurately can machine learning models predict slope failure and landslide risk by integrating multi-source geotechnical and rainfall datasets?

Background & Motivation: Slope failures and landslides are significant geotechnical hazards that lead to loss of life and infrastructure damage worldwide, especially in regions prone to heavy rainfall. Traditional predictive models often struggle with the complex, nonlinear interactions between soil properties, hydrology, and triggering events.

Research Gap / Question: While recent advances in machine learning offer promise for pattern recognition in large, heterogeneous datasets, there is limited research quantifying their prediction accuracy using real-world integrated geotechnical and meteorological data. The core question is whether these models can improve landslide risk forecasting over conventional methods.

Approach & Expected Contribution: The project will systematically review the literature, select relevant machine learning algorithms (e.g., Random Forest, Support Vector Machines, Neural Networks), and apply them to open geotechnical and rainfall datasets (such as the OpenLISEM or NASA TRMM data). Model performance will be evaluated using statistical metrics and compared to baseline geotechnical approaches.

Why It Matters: Improving predictive accuracy of slope failure risk supports resilient infrastructure planning, disaster prevention, and optimized resource allocation, addressing a critical need in civil and environmental engineering.