1st Edition
Explainable Machine Learning for Geospatial Data Analysis A Data Centric Approach
Explainable machine learning (XML), a subfield of AI, is focused on providing complex AI models that are understandable to humans. This book highlights and explains the details of machine learning models used in geospatial data analysis. It demonstrates the need for a data-centric explainable machine learning approach for obtaining new insights from geospatial data. It presents the opportunities, challenges, and gaps in the machine and deep learning approaches for geospatial data analysis and how they are applied to solve various environmental problems in land cover changes, and modelling forest canopy height and aboveground biomass. The author also includes guidelines and code scripts (R, Python) valuable for practical readers.
Features
- Includes data-centric explainable machine learning (ML) approaches for geospatial data analysis.
- Provides the foundations and approaches to explainable ML and deep learning.
- Includes several case studies from urban land cover and forestry where existing explainable machine learning methods are applied.
- Identifies opportunities, challenges and gaps in data-centric explainable ML approaches for geospatial data analysis.
- Provides scripts in R and python to perform geospatial data analysis.
This book is an essential resource for graduate students, researchers, and academics working and studying data science and machine learning. Geospatial data science professionals using GIS and remote sensing in environmental fields will also benefit from the new insights the author provides readers.
Part I: Introduction. 1. Machine Learning for Geospatial Data Analysis: Challenges and Opportunities. Part II: Foundations. 2. An Introduction to Explainable Machine Learning. 3. Approaches to Explainable Machine Learning. 4. Approaches to Explainable Deep Learning. 5. Landslide Susceptibility Modeling using a Logistic Regression Model. Part III: Techniques and Applications. 6. Urban Land Cover Classification using Earth Observation (EO) Data and Machine Learning Models. 7. Modeling Forest Canopy Height using Earth Observation (EO) Data and Machine Learning Models. 8. Modeling Aboveground Biomass Density using Earth Observation (EO) Data and Machine Learning Models. 9. Explainable Deep Learning for Mapping Building Footprints using High-Resolution Imagery. 10. Towards Explainable AI and Data-centric Approaches for Geospatial Data Analysis. 11. Appendix.
Biography
Courage Kamusoko is an independent geospatial consultant based in Japan. His expertise includes land use and land cover change modelling, and the design and implementation of geospatial database management systems. He is focused on the analyses of remotely sensed images and machine learning. He teaches practical machine learning for geospatial data analysis and modelling at the University of Tsukuba, Japan. Dr. Kamusoko has authored and co-edited six books total. He has contributed many chapters in published books along with conference papers and proceedings.
