Median control charts for monitoring asymmetric quality characteristics double bounded

Abstract

In many practical situations, the quality characteristics of interest assume values in the range (0,1), like rates and proportions (but they are not results from Bernoulli experiments). Most control charts built for these quality characteristics rely on monitoring parameters of their probability distribution functions or on their averages after some reparameterization of their density probability function. However, for highly asymmetric distributions, the median is a more appropriate location parameter than the average. In this paper, we propose Shewhart-type control charts for monitoring the median of observations taken from quality characteristics double bounded after reparameterization of two probability density functions: Kumaraswamy and unit Weibull. The performance of the control charts is evaluated and compared in terms of run length (RL) analysis considering three estimators for the median. Finally, we also carry out two applications to demonstrate the applicability of these control charts.

Control chart to monitor circular data

“…As considered in ref. [42] and due to the expensive computational time consumption of the resampling methods, we just consider the case‐K (where the bootstrap is replaced by Monte Carlo simulations) and the control limits determined by the same distribution considered in the data generation. For a fixed value, we consider the following algorithm to evaluate the circular control charts with subgroups: …”

A hybrid method based on P and P′ control chart for identifying hotspots

“…Du Nguyen and Phuc Tran 53 suggested two one‐sided control charts to detect the ratio between two normal random variables that play a significant role in various processes, the outcome showed that the advantages of the one‐sided control chart outweigh the two one‐sided charts in monitoring and perceiving any minute shift in a procedure in a factory or nonmanufacturing environment. Lima‐Filho et al 54 . developed a chart to trace the observations taken from quality features, which double bounded after reparameterization of the probability density functions.…”