Models economètrics en entorns big data enfocats a desigualtats

Perafita Basart, Xavier
In the last twenty years, the study of social inequalities has been a topic of special interest, revealing how differences between social classes have increased. During the same period, data generation has exponentially grown, leading to the normalization of the term "Big Data," which is widely recognized by the general population. This thesis examines the econometric mechanisms used in the management of massive data and applies them to social inequalities. To achieve this, existing mechanisms have been identified through a systematic review, which revealed five types of algorithms: classifiers, penalized classifiers, quasi-Bayesian penalized classifiers, clusters, and neural networks. Based on this review, the results of different algorithms have been utilized and compared to perform the clustering of municipalities in the province of Girona. The best grouping has served as the foundation for designing the sampling of the cohort driven by the Social, Environmental, and Health Observatory of Dipsalut. To study the inequality of vulnerable families in Catalonia, data on the typology of households and their environments have been obtained to analyze the living conditions of these families. The market they have access to is more limited, where the characteristics of the homes do not play a significant role in the possibility of renting housing; instead, the environment determines where they can live. These environments are more insecure, unhealthy, and impoverished, highlighting how low-income families are forced to reside in settings that exacerbate the poverty trap. Additionally, the study demonstrates that different logics regarding inequalities exist in urban and rural areas, which should be considered when designing and implementing any policies addressing inequality ​
​L'accés als continguts d'aquesta tesi queda condicionat a l'acceptació de les condicions d'ús establertes per la següent llicència Creative Commons: