Abstract

OSPINA HINCAPIE, CAROLINA; YANEZ CANAL, SERGIO  and  LOPERA GOMEZ, CARLOS MARIO. SOME BIVARIATE CONTROL CHARTS FOR ATRIBUTES. Dyna rev.fac.nac.minas [online]. 2010, vol.77, n.162, pp.325-337. ISSN 0012-7353.

Many industrial processes are multivariate in nature since the quality of a product depends on more than one variable. The multivariate control of processes captures the relation between the variables associated with the process, if this correlation is ignored and univariate control charts are used for every variable separately is possible to conclude erroneously over the process status. In the continuous case, many researches have been done, however there are few works that aim to correlated attributes. In this work we compare three charts of control for correlated bivariate binomial random variables, which are associated with attributes. The charts are: Hotelling's chart based on the approximation of the distribution binomial multivariate to the normal multivariate distribution. MNP chart which is an extension of univariate chart, and chart that is a non-parametric methodology based on the Mahalanobis's depth. The comparison is made through of simulation study using as a comparison measure the average run length (ARL). In this work we present an example of the used methodologies to construct control charts for bivariate binomial variables in a telecommunications company. The results shown in general terms that the MNP chart is the best in both control and out of control.

Keywords : Control Bivariate Charts; Hotelling's ; Bivariate Binomial Variables; MNP Control Chart; Control Chart; Mahalanobis's Depth; ARL.

        · abstract in Spanish     · text in Spanish     · Spanish ( pdf )