Detección de Comportamientos Atípicos de Contribuyentes con Riesgo de no Pago en una Administración Tributaria, Un Marco de Trabajo de Minería de Datos

Abstract

One of the primary processes in tax administration is debt collection management. The objective of this process, among others, is to recover economic resources that have been declared by taxpayers. Due to limitations in tax administration such as staffing, tools, time, and others, tax administrations seek to recover debts in the early stages of control where collection costs are lower than in subsequent stages. To optimize the debt collection management process and contribute to decision-making, this study proposes a deep learning-based framework to detect atypical behaviors of taxpayers with a high probability of non-payment. Normal and atypical behavior groups were also analyzed to identify interesting events using association rules.