On the Monitoring of Autocorrelated Linear Profiles

Control charts are commonly used to monitor quality of a process or product characterized by a quality characteristic or a vector of quality characteristics. However, in many practical situations the quality of a process or product can be characterized by a function or profile. Here we consider a linear function and investigate the violation of common independence assumption implicitly considered in most control charting applications. We specifically consider the case when profiles are not independent from each other over time. In this article, the effect of autocorrelation between profiles is investigated using average run length (ARL) criterion. Simulation results indicate significant impact on the ARL values when autocorrelation is overlooked. In addition, three methods based on time series approach are used to eliminate the effect of autocorrelation. Their performances are compared using ARL criterion.     

Statistical Monitoring of Autocorrelated Simple Linear Profiles Based on Principal Components Analysis

In this article, a transformation method using the principal component analysis approach is first applied to remove the existing autocorrelation within each profile in Phase I monitoring of autocorrelated simple linear profiles. This easy-to-use approach is independent of the autocorrelation coefficient. Moreover, since it is a model-free method, it can be used for Phase I monitoring procedures. Then, five control schemes are proposed to monitor the parameters of the profile with uncorrelated error terms. The performances of the proposed control charts are evaluated and are compared through simulation experiments based on different values of autocorrelation coefficient as well as different shift scenarios in the parameters of the profile in terms of probability of receiving an out-of-control signal.     

Phase I Analysis of Multiple Linear Regression Profiles

This article considers the Phase I analysis of data when the quality of a process or product is characterized by a multiple linear regression model. This is usually referred to as the analysis of linear profiles in the statistical quality control literature. The literature includes several approaches for the analysis of simple linear regression profiles. Little work, however, has been done in the analysis of multiple linear regression profiles. This article proposes a new approach for the analysis of Phase I multiple linear regression profiles. Using this approach, regardless of the number of explanatory variables used to describe it, the profile response is monitored using only three parameters, an intercept, a slope, and a variance. Using simulation, the performance of the proposed method is compared to that of the existing methods for monitoring multiple linear profiles data in terms of the probability of a signal. The advantage of the proposed method over the existing methods is greatly improved detection of changes in the process parameters of linear profiles with high-dimensional space. The article also proposes useful diagnostic aids based on F-statistics to help in identifying the source of profile variation and the locations of out-of-control samples. Finally, the use of multiple linear profile methods is illustrated by a data set from a calibration application at National Aeronautics and Space Administration (NASA) Langley Research Center.     

Phase II Monitoring of Geometric Profiles

Geometric profiles can be modeled effectively by large and small scale components. In several articles, a regression model with spatial autoregressive error term is combined with control charts to monitor geometric profiles. However, once a signal occurs, control charts would not be able to determine whether the shift is due to the large or small scale component. In this article, a combination of a multivariate and an omnibus control charts is used to monitor the large scale and small scale components to determine whether the shift is due to the large scale or small scale components.     

Phase I monitoring of social network with baseline periods using poisson regression

Social network analysis is an important analytic tool to forecast social trends by modeling and monitoring the interactions between network members. This paper proposes an extension of a statistical process control method to monitor social networks by determining the baseline periods when the reference network set is collected. We consider probability density profile (PDP) to identify baseline periods using Poisson regression to model the communications between members. Also, Hotelling T² and likelihood ratio test (LRT) statistics are developed to monitor the network in Phase I. The results based on signal probability indicate a satisfactory performance for the proposed method.     

Diagnosis Aids in Multivariate Multiple Linear Regression Profiles Monitoring

Diagnosis aids in addition to detecting the out-of-control state is an important issue in multivariate multiple linear regression profiles monitoring; because a large number of parameters and profiles in this structure are involved. In this paper, we specifically concentrate on identification of profile(s) and parameter(s) which have changed during the process in multivariate multiple linear regression profiles structure in Phase II. We demonstrate the effectiveness of our proposed approaches through Monte Carlo simulations and a real case study in terms of accuracy percent.     

Robust Estimation of Parameters in Simple Linear Profiles Using M-Estimators

In many situations, the quality of a process or product may be better characterized and summarized by a relationship between the response variable and one or more explanatory variables. Parameter estimation is the first step in constructing control charts. Outliers may hamper proper classical estimators and lead to incorrect conclusions. To remedy the problem of outliers, robust methods have been developed recently. In this article, a robust method is introduced for estimating the parameters of simple linear profiles. Two weight functions, Huber and Bisquare, are applied in the estimation algorithm. In addition, a method for robust estimation of the error terms variance is proposed. Simulation studies are done to investigate and evaluate the performance of the proposed estimator, as well as the classical one, in the presence and absence of outliers under different scenarios by the means of MSE criterion. The results reveal that the robust estimators proposed in this research perform as well as classical estimators in the absence of outliers and even considerably better when outliers exist. The maximum value of variance estimate in one scenario obtained from classical estimator is 10.9, while this value is 1.66 and 1.27 from proposed robust estimators when its actual value is 1.     

Change Point Estimation of Multivariate Linear Profiles Under Linear Drift

In this paper, maximum likelihood estimators (MLE) for both step and linear drift changes in the regression parameters of multivariate linear profiles are developed. Performance of the proposed estimators is compared under linear drift changes in the regression parameters when a combined MEWMA and Chi-square control charts method signals an out-of-control condition. The effect of smoothing parameter of MEWMA control charts, missing data, and multiple drift changes on the performance of the both estimators is also evaluated. The application of the proposed estimators is also investigated thorough a numerical example resulted from a real case.     

Phase I monitoring of generalized linear model-based regression profiles

In some industrial applications, the quality of a process or product is characterized by a relationship between the response variable and one or more independent variables which is called as profile. There are many approaches for monitoring different types of profiles in the literature. Most researchers assume that the response variable follows a normal distribution. However, this assumption may be violated in many cases. The most likely situation is when the response variable follows a distribution from generalized linear models (GLMs). For example, when the response variable is the number of defects in a certain area of a product, the observations follow Poisson distribution and ignoring this fact will cause misleading results. In this paper, three methods including a T²-based method, likelihood ratio test (LRT) method and F method are developed and modified in order to be applied in monitoring GLM regression profiles in Phase I. The performance of the proposed methods is analysed and compared for the special case that the response variable follows Poisson distribution. A simulation study is done regarding the probability of the signal criterion. Results show that the LRT method performs better than two other methods and the F method performs better than the T²-based method in detecting either small or large step shifts as well as drifts. Moreover, the F method performs better than the other two methods, and the LRT method performs poor in comparison with the F and T²-based methods in detecting outliers. A real case, in which the size and number of agglomerates ejected from a volcano in successive days form the GLM profile, is illustrated and the proposed methods are applied to determine whether the number of agglomerates of each size is under statistical control or not. Results showed that the proposed methods could handle the mentioned situation and distinguish the out-of-control conditions.     

Sequential Monitoring for Changes in Models with a Polynomial Trend

This article considers, for the first time, sequential monitoring procedures that detect possible parameter changes in the polynomial trend models. A new class of sequential monitoring procedures that based on generalized fluctuation testing principle is proposed. The asymptotic null distributions and the consistency of the proposed tests are derived and their asymptotic critical values are tabulated. The methods are illustrated and compared in a small simulation study. In particular, we apply the proposed tests to investigate the Chinese commodity retail sales and consumer price indexes. Simulations and applications support our method.     

Statistical Monitoring of Nominal Logistic Profiles in Phase II

Statistical analysis of profile monitoring, a relatively new sub-area of statistical process control due to its applications in different industries, have urged researchers and practitioners to contribute to the developments of new monitoring methods. A statistical profile is a relationship between a quality characteristic (a response) and one or more independent variables to characterize quality of a process or a product. In this article, statistical profiles based on nominal responses are studied, where logistic regression is used to model the responses. Three approaches including likelihood ratio test (LRT), multivariate exponentially weighted moving average (MEWMA), and support vector machines (SVM) approaches are proposed to monitor quality of a process or product in Phase II. Performances of the proposed approaches are evaluated and compared using a case study. Moreover, the effect of two important factors on average run length (ARL) performance, number of levels and number of covariates, has been considered. Results indicate that performance of all approaches depends on the number of covariates and levels. As the number of these factors increases, SVM performance improves while performance of the other approaches deteriorates.     

Phase I monitoring and change point estimation of autocorrelated poisson regression profiles

The objective of this paper is to study the Phase I monitoring and change point estimation of autocorrelated Poisson profiles where the response values within each profile are autocorrelated. Two charts, the SLRT and the Hotelling's T², are proposed along with an algorithm for parameter estimation. The detecting power of the proposed charts is compared using simulations in terms of the signal probability criterion. The performance of the SLRT method in estimating the change point in the regression parameters is also evaluated. Moreover, a real data example is presented to illustrate the application of the methods.     

Monitoring two-stage processes with a profile at the second stage

In the multistage processes, quality of a process or a product at each stage is related to the previous stage(s). This property is referred to as a cascade property. Sometimes, quality of a process is characterized by a profile. In this paper, we consider a two-stage process with a normal quality characteristic in the first stage and a simple linear regression profile in the second stage. Then we propose two methods to monitor quality characteristics in both stages. The performance of the proposed two methods is evaluated through a numerical example in terms of average run length criterion.     

A change point method for Phase II monitoring of generalized linear profiles

In this article, we adopt the change point approach to monitor the generalized linear profiles in phase II Statistical process control (SPC). Generalized linear profiles include a large class of profiles defined in one framework. In contrast to the conventional change point approach, we adopt the Rao score test rather than the likelihood ratio test. Simulated results show that our approach has a good performance over any possible single step change in process parameters for two special cases of generalized linear profiles, namely Poisson and binomial profiles. Some diagnostic aids are also given and a real example is introduced to shed light on the merits of our approach in real applications.     

Phase II Monitoring of Nonlinear Profiles

In many practical cases, the quality of a product or process is characterized by multiple measurements constituting a line or curve that is referred to as a profile. In this article, we develop two approaches for monitoring process and product nonlinear profiles. The first approach consists of control chart methods to monitor nonlinear profiles using parametric estimates of regression model. In order to avoid the problems arising from complexity of coefficient estimation of nonlinear profiles, the second approach, which consists of using metrics to measure deviation from a reference curve, is proposed. The performance of the methods is evaluated through a numerical example using average run length criterion. The effect of sample size on the performance of both approaches is also investigated in this article.     

The Performance of Phase II Simple Linear Profile Approaches when Parameters Are Estimated

Previous studies of statistical performance of Phase II simple linear profile approaches were reported only for the case of known profile parameters assumption. The main objective of this article is to evaluate and compare the performance of these approaches when the profile parameters are estimated from an in-control Phase I profile data set. Simulations establish that the performance of these approaches is strongly affected when the parameters are estimated compared to the known parameters case. The in-control performance of the competing approaches significantly deteriorates if estimated parameters are used with control limits intended for known parameters, especially when only a few Phase I samples are used to estimate the parameters. The results show also that some profile monitoring approaches need much larger number of Phase I profiles than other approaches to achieve the expected statistical performance. They also show that the profile monitoring approach proposed by Mahmoud et al. (2010 Mahmoud , M. A. , Morgan , J. P. , Woodall , W. H. ( 2010 ). The monitoring of simple linear regression profiles with two observations per sample . Journal of Applied Statistics 37 : 1249 – 1263 .[Taylor & Francis Online], [Web of Science ®] , [Google Scholar]) has generally better out-of-control run length performance than the competing approaches when the estimated parameters are used in the charts design.     

A Phase I/II adaptive design for heterogeneous groups with application to a stereotactic body radiation therapy trial

Dose‐finding studies that aim to evaluate the safety of single agents are becoming less common, and advances in clinical research have complicated the paradigm of dose finding in oncology. A class of more complex problems, such as targeted agents, combination therapies and stratification of patients by clinical or genetic characteristics, has created the need to adapt early‐phase trial design to the specific type of drug being investigated and the corresponding endpoints. In this article, we describe the implementation of an adaptive design based on a continual reassessment method for heterogeneous groups, modified to coincide with the objectives of a Phase I/II trial of stereotactic body radiation therapy in patients with painful osseous metastatic disease. Operating characteristics of the Institutional Review Board approved design are demonstrated under various possible true scenarios via simulation studies. Copyright © 2015 John Wiley & Sons, Ltd.     

A Three-Stage Bayesian Adaptive Phase I/II Design and Simulation Studies

Instead of using traditional separate phase I and II trials, in this article, we propose using a parallel three-stage phase I/II design, incorporating a dose expansion approach to flexibly evaluate the safety and efficacy of dose levels, and to select the optimal dose. In the proposed design, both the toxicity and efficacy responses are binary endpoints. A 3+3-based procedure is used for initial period of dose escalation at stage 1; at this level, the dose can be expanded to stage 2 for exploratory efficacy studies of phase IIa, while simultaneously, the safety testing can advance to a higher dose level. A beta-binomial model is used to model the efficacy responses. There are two placebo-controlled randomization interim monitoring analyses at stage 2 to select the promising doses to be recommended to stage 3 for further efficacy studies of phase IIb. An adaptive randomization approach is used to assign more patients to doses with higher efficacy levels at stage 3. We examine the properties of the proposed design through extensive simulation studies by using R programming language, and also compare the new design with the conventional design and a competing adaptive Bayesian design. The simulation results show that our design can efficiently assign more patients to doses with higher efficacy levels and is superior to the two competing designs in terms of total sample size reduction.     

Empirical Non-Parametric Control Charts: Estimation Effects and Corrections

Owing to the extreme quantiles involved, standard control charts are very sensitive to the effects of parameter estimation and non-normality. More general parametric charts have been devised to deal with the latter complication and corrections have been derived to compensate for the estimation step, both under normal and parametric models. The resulting procedures offer a satisfactory solution over a broad range of underlying distributions. However, situations do occur where even such a large model is inadequate and nothing remains but to consider non- parametric charts. In principle, these form ideal solutions, but the problem is that huge sample sizes are required for the estimation step. Otherwise the resulting stochastic error is so large that the chart is very unstable, a disadvantage that seems to outweigh the advantage of avoiding the model error from the parametric case. Here we analyse under what conditions non-parametric charts actually become feasible alternatives for their parametric counterparts. In particular, corrected versions are suggested for which a possible change point is reached at sample sizes that are markedly less huge (but still larger than the customary range). These corrections serve to control the behaviour during in-control (markedly wrong outcomes of the estimates only occur sufficiently rarely). The price for this protection will clearly be some loss of detection power during out-of-control. A change point comes in view as soon as this loss can be made sufficiently small.     

Exceedance probabilities for parametric control charts

Common control charts assume normality and known parameters. Quite often, these assumptions are not valid and large relative errors result in the usual performance characteristics such as the false alarm rate or the average run length. A fully nonparametric approach can form an attractive alternative but requires more Phase I observations than usually available. Sufficiently general parametric families then provide realistic intermediate models. In this article, the performance of charts based on such families is considered. Exceedance probabilities of the resulting stochastic performance characteristics during in-control are studied. Corrections are derived to ensure that such probabilities stay within prescribed bounds. Attention is also devoted to the impact of the corrections for an out-of-control process. Simulations are presented both to illustrate and to demonstrate that the approximations obtained are sufficiently accurate for practical usage.