Metodología de redes bayesianas para estimar el costo probable de proyectos de agua potable rural
Cargando...
Archivos
Fecha
2016
Autores
Profesor/a Guía
Facultad/escuela
Idioma
es
Título de la revista
ISSN de la revista
Título del volumen
Editor
Universidad Andrés Bello (Chile)
Nombre de Curso
Licencia CC
Licencia CC
Resumen
El presente estudio incursiona en las redes bayesianas aplicadas a proyectos de agua potable rural realizados en la Región de Los Ríos, Chile. Las redes bayesianas se caracterizan por ser grafos acíclicos dirigidos, es decir, relacionan variables representadas por nodos y permiten visualizar fácilmente las relaciones de dependencia entre ellas. La metodología considera, la identificación de las variables o nodos con sus respectivos estados, la configuración de la red piloto y a partir de los antecedentes de los proyectos de agua potable rural la construcción de las tablas de probabilidad condicional.
En términos de resultados, se trabajó con dos topologías, las cuales estaban conformadas por las ocho variables identificadas en los proyectos de agua potable rural. Se encontró que la topología número dos, proporcionaba mejores resultados en términos de predicción probabilística además de tener mejores estadísticos de evaluación como son pérdida logarítmica, compensación esférica, pérdida cuadrática.
Finalmente la validación y la evaluación de la red en términos de la predicción del costo probable de los proyectos del grupo de control. A partir de la metodología utilizada se observa una buena aproximación en términos de probabilidades para estimación de costos de este tipo de proyectos, aunque, el disponer de mayor número de datos incrementará el nivel de precisión.
The present study ventures into Bayesian networks applied to rural drinking water projects carried out in the Region of Rivers,Chile. Bayesian networks are characterized by being directed acyclic graph, that is, related variables represented by nodes and let you easily view the dependency relationships between them. The methodology considers the identification of variables or nodes with their respective states, the configuration of the pilot network and from the 2 history of the rural drinking water project construction of conditional probability tables. In terms of results, we worked with two typologies which were comprised of the eight variables identified in the rural drinking water projects. It was found that typology two provided better probabilistic forecasting results, as well as having better statistical evaluation capacity in areas such as are logarithmic loss, spherical compensation, and quadratic loss. Finally validation and evaluation of the network in terms of predicting the likely cost of the projects in the control group. From the methodology used a good approximation is observed in terms of probabilities to estimate costs of such projects, although, the availability of more data increases the level of accuracy.
The present study ventures into Bayesian networks applied to rural drinking water projects carried out in the Region of Rivers,Chile. Bayesian networks are characterized by being directed acyclic graph, that is, related variables represented by nodes and let you easily view the dependency relationships between them. The methodology considers the identification of variables or nodes with their respective states, the configuration of the pilot network and from the 2 history of the rural drinking water project construction of conditional probability tables. In terms of results, we worked with two typologies which were comprised of the eight variables identified in the rural drinking water projects. It was found that typology two provided better probabilistic forecasting results, as well as having better statistical evaluation capacity in areas such as are logarithmic loss, spherical compensation, and quadratic loss. Finally validation and evaluation of the network in terms of predicting the likely cost of the projects in the control group. From the methodology used a good approximation is observed in terms of probabilities to estimate costs of such projects, although, the availability of more data increases the level of accuracy.
Notas
Tesis (Ingeniero Civil)
Palabras clave
Abastecimiento de agua, Elaboración de proyectos, Chile