Modelo predictivo para riesgo de liquidez de una entidad fiduciaria usando minería de datos
Cargando...
Archivos
Autores
León Sánchez, Diana Patricia
Director
Tipo de contenido
Document language:
Español
Fecha
Título de la revista
ISSN de la revista
Título del volumen
Documentos PDF
Resumen
Este trabajo presenta un modelo predictivo para riesgo de liquidez basado en técnicas de minería de datos para una entidad fiduciaria. Se presentan los principales avances realizados desde la minería de datos en la gestión de riesgos financieros y otros métodos estadísticos empleados actualmente para riesgo de liquidez. El modelo en la fase de preprocesamiento analiza una serie histórica de registros de adiciones y retiros de una cartera colectiva y de otros factores macroeconómicos que inciden sobre el comportamiento del nivel de liquidez. Para la parte de predicción utiliza técnicas como redes neuronales y Naive Bayes. El modelo es validado con datos reales del producto fiduciario, los cuales son publicados por Asofiduciarias, Superfinanciera y la misma entidad.
Abstract This paper presents a predictive model for liquidity risk based on data mining techniques for trustee. Major progress since data mining in financial risk management and other statistical methods currently used for liquidity risk are presented. The model at the preprocessing phase analyzes a series of historical records additions and withdrawals from a pooled portfolio and other macroeconomic factors that in uence the behavior of liquidity. For the prediction part uses techniques such as neural networks and Nave Bayes. The model is validated with real data of the Trust product, which are published by Asofiduciarias, Superfinanciera and the same entity.
Abstract This paper presents a predictive model for liquidity risk based on data mining techniques for trustee. Major progress since data mining in financial risk management and other statistical methods currently used for liquidity risk are presented. The model at the preprocessing phase analyzes a series of historical records additions and withdrawals from a pooled portfolio and other macroeconomic factors that in uence the behavior of liquidity. For the prediction part uses techniques such as neural networks and Nave Bayes. The model is validated with real data of the Trust product, which are published by Asofiduciarias, Superfinanciera and the same entity.