Item
Universitat de Girona. Departament dâ€™InformÃ tica i MatemÃ tica Aplicada  
Vives Mestres, Marina
Daunis i Estadella, Josep MartÃn FernÃ¡ndez, Josep Antoni 

2011 May 11  
On standard control charts, the hypothesis of normality is usually assumed without any additional verification. Nevertheless, in some cases this assumption is not accurate and might cause errors in process quality monitoring. In particular, for the control of the proportion of nonconforming units (pchart) the normality is doubtful when p is small and consequently, lower control limit less than or equal to zero are frequent. Some authors have proposed new techniques to define limits in the pchart. Others have proposed transformations to improve the detection of special causes. In Xbar charts, the mean of a critical to quality (CTQ) characteristic is monitored. When the variable is far from normality, then parametric, or even nonparametric, control charts might be used. Recent works suggest applying transformations to make the data quasinormal. This kind of lack of normality is usually present when in those analyses CTQ is a part of a composition. Our proposal is to highlight how above mentioned problems can be treated from a compositional point of view. New strategies are proposed and illustrated trough a case study where a proportion has the role of a CTQ  
application/pdf  
http://hdl.handle.net/10256/13605  
eng  
Universitat de Girona. Departament dâ€™InformÃ tica i MatemÃ tica Aplicada  
CoDaWork 2011. The 4th International Workshop on Compositional Data Analysis  
Tots els drets reservats  
AnÃ lisi multivariable  Congressos
Multivariate analysis  Congresses EstadÃstica matemÃ tica  Congressos Mathematical statistics  Congresses AnÃ lisi dâ€™error (MatemÃ tica)  Congressos Error analysis (Mathematics)  Congresses 

Could CODA Methodology be Useful in Control Chart Techniques?  
info:eurepo/semantics/conferenceObject  
DUGiDocs 