Deep Learning for Next-Generation Geodemographic Classification | Alex Singleton | Uni. of Liverpool

In this session from the Spatial Data Science Conference New York 2025 (#SDSC25), Alex Singleton (Professor of Geographic Information Science, University of Liverpool) introduces a deep learning method for geodemographic classification using autoencoders. He discusses Residential Area Encoding, a technique that better captures complex geographic patterns than traditional approaches. 00:10 Introduction & background 01:51 What are geodemographics? 05:22 How classifications are built (inputs, clustering, labeling) 06:42 Key challenges & why geography complicates modeling 11:06 Introducing Residential Area Encoding (autoencoders) 15:55 Results and conclusions