Search published articles


Showing 4 results for Likelihood Ratio

Ebrahim Mazrae Farahani, Reza Baradaran Kazemzade, Amir Albadvi, Babak Teimourpour,
Volume 29, Issue 3 (9-2018)
Abstract

Studying the social networks plays a significant role in everyone’s life. Recent studies show the importance and increasing interests in the subject by modeling and monitoring the communications between the network members as longitudinal data. Typically, the tendency for modeling the social networks with considering the dependency of an outcome variable on the covariates is growing recently. However, these studies fail in considering the possible correlation between the responses in the modeling of social networks. Our study use generalized linear mixed models (GLMMs) (also referred to as random effects models) to model the social network according to the attributes of nodes in which the nodes take a role of random effect or hidden effect in the modeling. The likelihood ratio test (LRT) statistics is implemented to detect change points in the simulated network streams. Also, in the simulation studies, we applied root mean square Error (RMSE) and standard deviation criteria for choosing an appropriate distribution for simulation data. Also, our simulation studies demonstrates an improvement in the average run length (ARL) index in comparison to the previous studies.
 
Ahmad Hakimi, Hiwa Farughi, Amirhossein Amiri, Jamal Arkat,
Volume 33, Issue 1 (3-2022)
Abstract

In some statistical processes monitoring (SPM) applications, relationship between two or more ordinal factors is shown by an ordinal contingency table (OCT) and it is described by the ordinal Log-linear model (OLLM). Newton-Raphson algorithm methods have also been used to estimate the parameters of the log-linear model. In this paper, the OLLM based processes is monitored using MR and likelihood ratio test (LRT) approaches in Phase I. Some simulation studies are applied to performance evaluation of the proposed approaches in terms of probability of signal under step shifts, drifts and the presence of outliers. Results show that, by imposing the small and moderate shifts in the ordinal log-linear model parameters, the MR statistic has better performance than LRT. In addition, a real case study in dissolution testing in pharmaceutical industry is employed to show the application of the proposed control charts in Phase I.  

Mohammad Yaseliani, Majid Khedmati,
Volume 34, Issue 1 (3-2023)
Abstract

Diagnosis of diseases is a critical problem that can help for more accurate decision-making regarding the patients’ health and required treatments. Machine learning is a solution to detect and understand the symptoms related to heart disease. In this paper, a logistic regression model is proposed to predict heart disease based on a dataset with 299 people and 13 variables and to evaluate the impact of different predictors on the outcome. In this regard, at first, the effect of each predictor on the precise prediction of the outcome has been evaluated and analyzed by statistical measurements such as AIC scores and p-values. The logit models of different predictors have also been analyzed and compared to select the predictors with the highest impact on heart disease. Then, the combined model that best fits the dataset has been determined using two statistical approaches. Based on the results, the proposed model predicts heart disease with a sensitivity and specificity of 84.21% and 90.38%, respectively. Finally, using normal probability density curves, the likelihood ratios have been established based on classes 1 and 0. The results show that the likelihood ratio classifier performs as satisfactorily as the logistic regression model.
Ali Salmasnia, Mohammad Reza Maleki, Esmaeil Safikhani,
Volume 34, Issue 2 (6-2023)
Abstract

In some applications, the number of quality characteristics is larger than the number of observations within subgroups. Common multivariate control charts to monitor the variability of such high-dimensional processes are unsuitable because the sample covariance matrix is not positive semi-definite and invertible. Moreover, the impact of gauge imprecision on detection capability of multivariate control charts under high-dimensional setting has been clearly neglected in the literature. To overcome these shortcomings, this paper develops a ridge penalized likelihood ratio chart for Phase II monitoring of high-dimensional process in the presence of measurement system errors. The developed control chart departures from the assumption of sparse variability shifts in which the assignable cause can only affects a few elements of the covariance matrix. Then, to compensate for the adverse impact of gauge impression, the developed chart is extended by employing multiple measurements on each sampled item. Simulation studies are carried out to study the impact of imprecise measurements on detectability of the developed monitoring scheme under different shift patterns. The results show that the gauge inability negatively affects the run-length distribution of the developed control chart. It is also found that the extended chart under multiple measurements strategy can effectively reduce the error impact.

Page 1 from 1