Analysing and measuring the risk of eviction with multimodal machine learning

• Grant number: 24/12863-8
• Support Opportunities: Scholarships in Brazil - Doctorate (Direct)
• Start date: December 01, 2024
• End date: August 31, 2029
• Field of knowledge: Interdisciplinary Subjects
• Principal Investigator: Fabio Kon
• Grantee: Pedro Henrique Rezende Mendonça
• Host Institution: Instituto de Matemática e Estatística (IME). Universidade de São Paulo (USP). São Paulo , SP, Brazil
• Associated research grant: 23/00811-0 - EcoSustain: computer and data science for the environment, AP.TEM

Abstract

Processes of evictions are characterized by the forced displacement of people from their houses. Obtaining information about these processes poses a research challenge, often tackled qualitatively. Identifying them on a broader scale and thus understanding their effects in large urban centers requires employing other methods and technologies. The diversity of data sources about cities offers a heterogeneous description of urban dynamics, but not specifically about evictions. This diversity imposes methodological challenges for broadening the spatiotemporal coverage of mapping and monitoring the risk of eviction. This Doctorate project aims to apply Data Science methods, such as modeling and machine learning with spatial and non-spatial data, to enable frequent, quantitative, and comprehensive identification and monitoring of the risk of eviction. A case study will be conducted in the context of the São Paulo Metropolitan Region, starting with the mapping carried out by the Observatory of Evictions (OR) of the Faculty of Architecture and Urbanism of the University of São Paulo. The modeling of the problem and the implementation of a computational solution will allow for a comparative evaluation of the performance of unimodal and multimodal learning algorithms with spatial databases. The results will be consolidated into open software and made available with data visualization tools for broad access aimed at a technical audience. This is an interdisciplinary project involving the areas of Data Science, Computer Science, and Urban Planning.