In this paper, ranked set sampling is proposed for constructing a non-parametric quality control chart based on sign statistic for monitoring the process center when the units are difficult and costly to measure. The characteristics of run length such as average, median and standard deviation are used as criteria to compare two sampling schemes: ranked set sampling and simple random sampling. Cauchy, double exponential and normal distributions are applied to investigate the performance of sign control chart based on studied sampling schemes. The proposed control chart is illustrated with a real data set. The results indicate that when the process starts to get out of control, the chart based on ranked set sampling reduces the run length characteristics values.
Unlike continuous distributions, discrete models capable of modeling failure times with bathtub-shaped hazard rate functions have not received deserving attention in the statistical reliability literature. In this article, we show that the shape of the hazard rate function of a continuous lifetime model is partially preserved under the usual discretization of the model. Then, two continuous distributions with bathtub-shaped hazard rates are discretized, and some properties of the obtained discrete versions are studied. Graphical methods are utilized to obtain initial estimates of the parameters that are necessary for calculation of the maximum likelihood estimator. Finally, we fit the studied models along with some other proper discrete models to a set of real data and show that these models may be good candidates for describing discrete lifetime data.
31 July 2018 in Physica A: Statistical Mechanics and its Applications