Conference Papers

Permanent URI for this collectionhttps://idr.nitk.ac.in/handle/123456789/28506

Browse

Author: search.filters.author.Kundapura, S.Subject: search.filters.subject.Machine Learning

Search Results

Now showing 1 - 3 of 3
  • No Thumbnail Available
    Item
    Machine Learning Ensemble Model for Flood Susceptibility Mapping
    (Institute of Electrical and Electronics Engineers Inc., 2022) Kundapura, S.; Soman, A.; Kuruvilla, E.
    Floods can be considered the most dangerous natural disaster, given their unpredictability and capacity to wipe out valuable life and property. The timely and efficient prediction of floods and flood susceptible or risk zones has been of utmost importance and can help with risk assessment, long-term management, and future preparedness. Over the years, Machine Learning (ML) has evolved as a powerful tool to build accurate flood models, inundation maps, and warning systems. This study uses the Support Vector Machine (SVM) and Random Forest (RF) algorithms and a Bagging Ensemble Model for Flood Susceptibility Mapping of Ernakulam, Kerala, India using multi-source geospatial data and the WEKA software. A total of twelve Flood Conditioning Factors (FCFs) are considered as variables, namely elevation, slope, curvature, Topographic Roughness Index (TRI), Topographic Wetness Index (TWI), Stream Power Index (SPI), rainfall, Land Use Land Cover (LULC), distance to the river, drainage density, and geology. The contribution of factors is assessed using the OneR Feature Selection method. The models are compared using the Receiver Operating Characteristics (ROC) curve and the Area Under the Curve (AUC) method. The Flood Susceptibility Map provides an opportunity for planners and authorities to flood preparation and long-term planning against flood impacts. © 2022 IEEE.
  • No Thumbnail Available
    Item
    Performance Comparison of Machine Learning Algorithms in Groundwater Potability Prediction
    (Institute of Electrical and Electronics Engineers Inc., 2022) Kuruvilla, E.; Kundapura, S.
    Rising global water demand has resulted in the overuse of groundwater resources and a decline in groundwater quality. Physical and chemical characteristics significantly impacted by geological formations and human activities determine how groundwater quality varies. An accurate and reliable assessment of groundwater resource information is the key element for effective management and enhancement of groundwater quality. The utilization of modern Machine Learning (ML) techniques in groundwater quality assessment provides insights for policymakers in suggesting remedies and management approaches for groundwater quality issues. Machine Learning models outperform other simulation models, using input and output datasets without considering the intricate relationship of the model to be analyzed and decreasing computational efforts. Comparison of various ML techniques, including Ensemble, Nonlinear, and Linear models for the prediction of groundwater potability is the main objective of this study. The presence of potable groundwater suggests that the water is fit for human consumption. The proposed approach makes use of eight ML algorithms i.e. Gradient Boosting Classifier (GB), Random Forest (RF), Decision Tree (DT), K-Nearest Neighbors (KNN), Naïve Bayes (NB), Support Vector Machine (SVM), Linear Regression (LR) and Stochastic Gradient Descent (SGD) algorithm. According to the results, the Ensemble ML models outperformed well followed by the Nonlinear models, and Linear classification ML models have comparatively less accuracy and reliability. © 2022 IEEE.
  • No Thumbnail Available
    Item
    AN Integrated Analysis and Forecasting of Wildfires in the Nallamala Hills, India
    (Institute of Electrical and Electronics Engineers Inc., 2023) Kundapura, S.; Vishnu Vardhan, M.; Apoorva, K.V.
    Wildfires threaten ecosystems, human lives, and infrastructure, necessitating effective detection and prediction methods. In this study, an in-depth analysis of wildfire detection and forecasting is carried out over the Nallamala hills, which stretch across the states of Telangana and Andhra Pradesh. Our approach comprises three significant steps: Active fire analysis, pre-fire analysis, and post-fire analysis. Pre-fire maps were created using the Normalised Difference Vegetation Index (NDVI) during the pre-fire analysis, which involved time series analysis of significant components. For active fire analysis, the first dataset is created by using satellite imagery and its derived products. A dataset is used to train the five different machine-learning models for prediction. Among these models, the Random Forest classifier outperformed the remaining four models (Support vector Classifier, Gradient Boosting Classifier, Logistic Regression, and K-means algorithms) in accurately detecting and predicting active fires. This step enabled real-Time monitoring and prioritisation of firefighting efforts. The burnt area calculation uses the Normalised Burn Ratio (NBR) in the post-fire analysis. The analysis implemented post-fire rehabilitation and restoration efforts, giving essential information on the scope and severity of fire damage. The comprehensive study of all wildfires will provide a detailed picture of what occurred in the past (Timeseries), present (Prediction models), and future (Pre-fire maps), allowing people and government agencies to take precautions against future wildfires. © 2023 IEEE.