What factors contribute to uneven suburbanisation? Predicting the number of migrants from Warsaw to its suburbs with machine learning

Abstract This article investigates the spatially uneven migration from Warsaw to its suburban municipalities. We report a novel approach to modelling suburbanisation: several linear and nonlinear predictive models are applied, and Explainable Artificial Intelligence methods are used to interpret the shape of relationships between the dependent variable and the most important regressors. The support vector regression algorithm is found to yield the most accurate predictions of the number of migrants to the suburbs of Warsaw. In addition, we find that migrants choose wealthier and more urbanised municipalities that offer better institutional amenities and a shorter driving time to Warsaw’s city centre.