Reglas de clasificación discriminante: aplicación a la vivienda

In the Encuesta de Presupuestos Familiares, we find data for many households missing any information about the legal status of their housing. In this paper, we use the Logistic Regression Analysis method to define four rules allowing to estimate the classification: "free price" and "projected price". Two types of error may arise: i) type-I error, free allocation when it is protected, and ii) type-II error, for the opposite case. While the classification rule that directly arises from the logistic estimation is based on minimizing the sum of both errors, our rules follow a criterion that allows to apply a stronger penalty against the type-I error. The difference between the rules relies on the trade-off between the number of classified habitations and the associated error. The use of this technique allows to enlarge the information available in the Encuesta de Presupuestos Familiares in order to estimate either the evolution or the determinants of prices according to this database.