Most businesses sit on a goldmine of geolocation data, from tracking fleet movements to optimizing supply-chain networks. However, the true value isn't in seeing dots on a map; it's in uncovering the hidden patterns within the movement. This session dives into the technical lifecycle of Geospatial AI, moving beyond basic visualization into the world of Unsupervised Machine Learning and Spatial Graphs.

We will start by building a robust data foundation, walking through a high-performance pipeline using Python (GeoPandas) and SQL (PostGIS) to process GPS signals and handle spatial joins.

From there, we dive into:

• Pattern Discovery: Using unsupervised models like DBSCAN and K-Means to segment movement types and identify behavioral clusters.
• The Power of Graphs: How geospatial graphs can be a game-changer for Supply/Demand Optimization.
• The Frontiers of Geo-AI: A brief look at the intersection of Graph Neural Networks and how Generative AI is poised to revolutionize how geospatial data is used, moving toward a future where models autonomously interpret complex spatial data models.

By the end of this talk, you’ll understand how to transform raw GPS traces into a sophisticated AI pipeline capable of driving strategic business decisions.

Matthias Rochtus, experienced public speaker and Data Scientist.
Main fields of expertise are Geospatial research, data engineering & CI/CD , Predictive Maintenance and Retrieval Augmented Generation (LLMs)