The aim of this work is to present a new methodology to measure the quality of a service. A nonparametric model is developed in which customers evaluate the overall service quality and a set of dimensions or attributes that determine this service quality. The model assumes that overall service quality is determined by a linear combination of attributes evaluations with some unknown weights and that different customers may have different weights for the attributes. The nonparametric techniques are based in Nearest Neighbours combined with Restricted Least Squared methods. The model is applied to several simulated data sets where we know the true value of the parameters of the model. Then we have applied the methodology to a specific set of data from CABINTEC ("Intelligent cabin truck for road transport"). Finally, the methodology is applied to the measurement of the quality of the postgraduate courses of a public Spanish University. The methodology, that we call ALR Adaptive Local Regression, have demonstrate be able to treat these kind of data. ALR permits to calculate the weight that customer assignsto each quality attribute of the service.

Tesis Doctoral leída en la Universidad Rey Juan Carlos de Madrid en 2010. Directores de la Tesis: Dr. Javier Martínez Moguerza y Dr. Clara laura Cardone Riportella

Estadística e Investigación Operativa