Utilizing AutoML in ArcGIS for Geospatial Machine Learning
Samuel Rivera
Co-Presenters: Individual Presentation
College: The Dorothy and George Hennings College of Science, Mathematics and Technology
Major: Information Technology
Faculty Research Mentor: Daehan Kwak
Abstract:
ArcGIS is an application that allows the user to explore, analyze, and visualize data. Within this application, there are multiple toolboxes that are available for the user, one being AutoML. This project explores the AutoML tool within ArcGIS to predict the elevation of locations based on their geospatial coordinates using machine learning models. The dataset with known elevation points was used to train various machine learning models within the AutoML tool in ArcGIS. The goal was to predict the elevation of a given location based on its longitude and latitude. During the training process, AutoML selected optimal features, tuned hyperparameters, and evaluated multiple models, including Random Forest, Decision Tree, Linear, Xgboost, LightGBM, which enables efficient and accurate elevation predictions. The tool then gave a summary report of the training data and showed the best model with explanation through various statistics. Unseen testing data was then fed to the best performing model, and it resulted in predictions of elevation in meters. The performance of the model created was assessed using a python expression to categorize the results to calculate accuracy. Additional datasets will be used to further explore and enhance machine learning applications in ArcGIS.