Regression Models and Machine Learning for Predicting Attractive Cities for Talent

Share

Abstract

This Bachelor’s Thesis addresses the quantitative analysis and prediction of urban attractiveness across 175 global cities in the face of increasing competition for talent and investment. Using a socioeconomic indicators database from the Observatorio UFV de Ciudades Atractivas para el Talento, the report details the design and implementation of Decision Support System (DSS) based on Mathematical Engineering techniques. Following the CRISP-DM methodology, the project covers everything from data clearing and unsupervised segmentation via K-Means to the comparative training of predictive models (Linear Regression, Random Forest, XGBoost and Neural Networks) to estimate urban attractiveness, magnetism and profitability. Additionally, Explainable Artificial Intelligence (SHAP) is integrated to identify the causal drivers behind the predictions. The document details the technical experimentation, the development of a projection and stress-testing simulator (What-If analysis) and evaluates the ethical, legal and social impact of the solution, ensuring a transparent scientific framework to serve as a strategic tool for public planning and investment.