Cloud Native Geospatial Analytics with Apache Sedona

Chapter 7. Geospatial Data Science and Machine Learning

Geospatial data science is an interdisciplinary field that uses geospatial data (raster and vector data) with machine learning techniques to understand the surrounding world and support decision making in various industries, such as transportation, climate, retail, and real estate.

Many algorithms were invented in the mid-20th century, but due to the lack of efficient and cheap computing power, they were impossible to apply to business-scale problems. With simplified access to efficient computer power in recent years, such as cloud providers, we can solve many previously unsolvable issues. We can apply many machine learning models to improve our daily lives. That is what Apache Sedona is aiming for as well, to make solving complex geospatial problems with the use of statistics and machine learning models easily accessible.

This chapter will teach you how to use Apache Sedona with geospatial statistics algorithms like Moran’s I and local outlier detection. We will walk you through applying machine learning models like DBSCAN, KMeans, and XGBoost to solve classification and clustering problems, integrating your Apache Sedona application with MLlib. Raster data is semi-structured data containing a lot of information. One example is combining an image segmentation model to create vector data from raster data in Apache Sedona.

Geospatial Clustering with Apache Sedona (DBSCAN)

Density-based spatial clustering of applications with ...