Hunting for bubbles: A predictive model of New York city's real estate market

• Autor(a) principal: Soares, Nuno Rodrigo Basílio
• Data de Publicação: 2023
• Tipo de documento: Dissertação
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: http://hdl.handle.net/10071/30055
• topic: Housing bubbles; Housing market; Machine learning; Classification models; Economic drivers; XGBoost; Bolhas imobiliárias; Mercado imobiliário; Modelos de classificação; Impulsionadores económicos

What defines a housing bubble precisely? What forms the consensus on its principal triggering factors? Are these factors relevant to New York City? Housing bubbles are not novel occurrences, one was even associated with the major financial crisis of 2008. Understanding the causal factors behind a housing bubble and the unique dynamics characterizing the renowned New York City real estate market, are subjects frequently discussed by both the public and the media. The aim of this dissertation is to examine the primary economic factors typically linked to housing bubbles, drawing insights from research conducted in various geographical regions. This research involves gathering data to monitor these economic drivers in the US and New York City markets. Subsequently, classification models were constructed using these variables as inputs to identify periods of housing bubbles. The dissertation also includes an analysis of model performance and a comparison between different models to determine which feature set yields superior results. This study concluded that the most successful model configuration was achieved by using XGBoost with the feature list of Test 1. This configuration was tested both with and without feature selection, resulting in accuracy rates of 0.89 and 0.86, respectively. Notably, the features that played a significant role in our classification align with those highlighted by other researchers as crucial for housing bubble detection. Features such as the price-to-rent ratio, inflation, and interest rates demonstrated their applicability to New York City, substantiating findings from diverse geographical regions.