Phase II monitoring of autocorrelated attributed social networks based on generalized estimating equations

Hatef Fotuhi | Amirhossein Amiri | Ali Reza Taheriyoun

---

URL :   http://research.shahed.ac.ir/WSR/WebPages/Report/PaperView.aspx?PaperID=158388
Date :  2021/03/04
Publish in :    Communications in Statistics Part B: Simulation and Computation
DOI :  https://doi.org/10.1080/03610918.2021.1887224
Link :  http://dx.doi.org/10.1080/03610918.2021.1887224
Keywords :Average run length; Generalized estimating equations; Generalized Linear model; Phase II; Social network monitoring

Abstract :

---

Social network monitoring has attracted the attention of many researchers in the area of statistical process monitoring. It is essential to consider the categorical attributes of communications between people in the modeling of social networks. Social network communications with categorical attributes can be summarized in contingency tables and it is more realistic to work with the communications as time-dependent variables. In the present study, since the autocorrelation between social networks can significantly alter the statistical performance of different monitoring approaches, generalized estimating equations are used to estimate model parameters considering autocorrelation between social networks. Moreover, two control charts, Hotelling’s T2 and the multivariate exponentially weighted moving average, are developed to monitor model parameters in Phase II. The performance of the proposed method is evaluated through simulation studies in terms of the average run length criterion. Finally, the application of the proposed method is shown through a numerical example and a real case.