BAYESIAN CONTROL CHARTS FOR ATTRIBUTES AND THE LARGE SUBGROUP SIZE IN p CHART CARTAS DE CONTROL BAYESIANAS PARA ATRIBUTOS Y EL TAMAÑO DE SUBGRUPO GRANDE EN LA CARTA p

The problems of the control charts for attributes are analyzed. In particular, it is seen that the traditional procedure to obtain the control limits has several difficulties: it does not incorporate the uncertainty in the estimate of the parameter of the distribution, it does not include the variat...

Full description

Bibliographic Details
Main Author: Gutiérrez Humberto
Format: Article
Language:English
Published: Universidad Nacional de Colombia 2006-11-01
Series:Revista Colombiana de Estadística
Subjects:
Online Access:http://www.revistas.unal.edu.co/index.php/estad/article/view/29180
Description
Summary:The problems of the control charts for attributes are analyzed. In particular, it is seen that the traditional procedure to obtain the control limits has several difficulties: it does not incorporate the uncertainty in the estimate of the parameter of the distribution, it does not include the variation of the process, and the procedure requires a base period to obtain the control limits. Along this period it does not have a control chart to analyze the process. It is seen that these problems are solved applying Bayesian methods. It is described a sequential procedure to obtain the Bayesian control limits for p, np, c and u charts, from the first observation. This eliminates the necessity of the base period. Lastly it is analyzed the problem of the p chart when it is based on a large subgroup size and it is proposed a Bayesian solution for this problem.<br>Se analizan los problemas de las cartas de control para atributos. En particular, se ve que el procedimiento tradicional para obtener los límites de control tiene varias dificultades: no incorpora la incertidumbre sobre la estimación del parámetro del modelo, no contempla las variaciones en el parámetro del proceso y requiere un período base para obtener datos. Durante este período base no se tiene una carta de control para monitorear el proceso. Se ve que estos problemas son resueltos con métodos Bayesianos. Se ve cómo establecer los límites de control Bayesianos para las cartas de control p, np, c y u, en forma secuencial desde la primera observación. Esto elimina la necesidad del período base. Por último se analiza el problema de la carta p cuando esta se basa en un tamaño de subgrupo grande y se propone una solución Bayesiana para este problema.
ISSN:0120-1751