نوع مقاله : پژوهشی
نویسندگان
دانشکده مهندسی صنایع-دانشگاه صنعتی خواجه نصیرالدین طوسی
چکیده
کلیدواژهها
عنوان مقاله [English]
نویسندگان [English]
This paper proposes a cumulative sum (CUSUM) control chart for monitoringcustomer satisfaction and loyalty. This control chart is developed on the basis of ordered categorical observations for customer satisfaction and loyalty which can be assessed regularly by means of customer satisfaction surveys. In calculating control statistic as well as setting control limit, customer-specific characteristics are taken into account as an observable covariate while last information of customer behavior is used to update the distribution of the unobservable part of customer attitude in a Bayesian manner. To account for the time varying functional relationship between customer satisfaction and loyalty, first a Markov chain model is used to indicate how customers past perception in terms of his satisfaction level may
influence his future attitude and then we use a logistic regression model to describe how customers loyalty level is affected by his satisfaction level. Explaining that customer satisfaction may have long but diminishing effect on
customer loyalty, the Markov chain model has preceded the logistic regression model that relates satisfaction to loyalty. The parameters of the regression model between satisfaction and loyalty provide a suitable ground for designing CUSUM control statistic. To calculate an approximate control limit for the CUSUM control chart, a Markov chain approach is adopted enabling us to obtain a pre-specified in-control average run length, given customers particular
characteristics that include both observable and unobservable parts. It is worth mentioning that a beta distribution serves as the prior distribution for customers unknown part of information whose distribution can be updated using
last observations regarding customer satisfaction and loyalty via customer satisfaction survey. Results of investigating the performance of CUSUM control chart in out-of-control situations where the parameters of the functional relationship vary are also analyzed to evaluate the detect ability of the proposed monitoring approach with respect to the changes in customer attitude.
کلیدواژهها [English]