Kim Phuc Tran

Monitoring the Ratio of two Normal Variables using Run Rules Type Control Charts

Recent studies show that Shewhart-type control charts monitoring the ratio of two normal random variables are useful to perform continuous surveillance in several manufacturing environments; anyway, they have a poor statistical sensitivity in the detection of small or moderate process shifts. The statistical sensitivity of a Shewhart control chart can be improved by implementing supplementary Run Rules. In this paper, we investigate the performance of Phase II Run Rules Shewhart control charts monitoring the ratio with each subgroup consisting of sample units. A Markov chain methodology coupled with an efficient normal approximation of the ratio distribution is used to evaluate the statistical performance of these charts. We provide an extensive numerical analysis consisting of several tables and figures to discuss the statistical performance of the investigated charts for deterministic and random shift sizes affecting the in-control ratio. An illustrative example from the food industry is provided for illustration.

The performance of the Shewhart-RZ control chart in the presence of measurement error

Recent literature about quality control has investigated the continuous surveillance of the ratio of two normal random variables under the assumption of no measurement error. However, in practice, measurement errors always exist in quality control applications and may considerably affect the performance of control charts. In this paper, the performance of the Shewhart-RZ control chart is investigated in the presence of a measurement error and modelled by a linear covariate error model. Several figures and tables are generated and commented to show the statistical performance of the Shewhart-RZ control chart for different sources of the measurement error. Two examples illustrate the use of this chart on a quality control problem simulated from the food industry and a real industrial case from a plant treating batteries for recyclement.

On the Performance of Shewhart median Chart in the Presence of Measurement Errors

In the literature, median control charts have been introduced under the assumption of no measurement error. However, measurement errors always exist in practice and may considerably affect the ability of control charts to detect out-of-control situations. In this paper, we investigate the performance of Shewhart median chart by using a linear covariate error model. Several figures and tables are presented and commented to show the statistical performance of Shewhart median control chart in the presence of measurement errors. We also investigate the positive effect of taking multiple measurements for each item in a subgroup on the performance of Shewhart median chart. An example illustrates the use of Shewhart median chart in the presence of measurement errors. Copyright

The Efficiency of the 4-Out-of-5 Runs Rules Scheme for Monitoring the Ratio of Population Means of a Bivariate Normal Distribution

In order to monitor the ratio of population means of a bivariate normal distribution, [K. P. Tran, P. Castagliola, and G. Celano, Monitoring the ratio of two normal variables using run rules type control charts, Int. J. Prod. Res. 54(6) (2016a) 1670–1688] recently suggested a scheme based on Runs Rules, called the RRRZ control chart, and showed how their proposed approach efficiently detects small or moderate process shifts. The goal of this paper is to investigate the one-sided RRRZ control charts with the 4-out-of-5 Runs Rules by using a Markov chain methodology. Several tables are generated and commented to show the statistical performance of the investigated charts. Also, comparisons with other competitive schemes are provided to show the statistical performance of the proposed charts. Finally, two examples illustrate the use of proposed charts on a quality control problem simulated from the food industry and a real industrial case from a plant treating batteries for recyclement.

Run Rules median control charts for monitoring process mean in manufacturing

Recent studies show that Shewhart median ( X˜) chart is simpler than the Shewhart X¯ chart and it is robust against outliers, but it is often rather inefficient in detecting small or moderate process shifts. The statistical sensitivity of a Shewhart control chart can be improved by using supplementary Run Rules. In this paper, we propose the Phase II median Run Rules type control charts. A Markov chain methodology is used to evaluate the statistical performance of these charts. Moreover, the performance of proposed charts is investigated in the presence of a measurement errors and modelled by a linear covariate error model. An extensive numerical analysis with several tables and figures to show the statistical performance of the investigated charts is provided for both cases of measurement errors and no measurement errors. An example illustrates the use of these charts.

Data driven hyperparameter optimization of one-class support vector machines for anomaly detection in wireless sensor networks

One-class support vector machines (OCSVM) have been recently applied to detect anomalies in wireless sensor networks (WSNs). Typically, OCSVM is kernelized by radial bais functions (RBF, or Gausian kernel) whereas selecting Gaussian kernel hyperparameter is based upon availability of anomalies, which is rarely applicable in practice. This article investigates the application of OCSVM to detect anomalies in WSNs with data-driven hyperparameter optimization. Specifically, the information of the farthest and the nearest neighbors of each sample is used to construct the objective cost instead of labeling based metrics such as geometric mean accuracy (G-mean) or area under the receiver operating characteristic (AUROC). The efficiency of this method is illustrated over the IBRL dataset whereas the resulting estimated boundary as well as anomaly detection performance are comparable with existing methods.

A Variable Sampling Interval EWMA Distribution-Free Control Chart for Monitoring Services Quality

Many data in service quality came from a nonnormal or unknown distribution, hence the commonly-used control charts are not suitable. In this paper, new Arcsine Shewhart Sign and Variable Sampling Interval EWMA (Exponentially Weighted Moving Average) distribution-free control charts are proposed. The procedure does not require the assumption of normal data. A Markov chain method is used to obtain optimal designs and evaluate the statistical performance of the proposed charts. Furthermore, practical guidelines and comparisons with the basic Arcsine EWMA Sign control chart are provided. Results show that the proposed chart is considerably more efficient than the basic Arcsine EWMA Sign control chart. The proposed control charts are illustrated by analysing the service quality of the Vancouver City Call Centre.

Real Time Data-Driven Approaches for Credit Card Fraud Detection

Credit card fraud causes many financial losses for customer and also for the organization. For this reason, in the past few years, many studies have been performed using machine learning techniques to detect and block fraudulent transactions. This paper introduces two real time data-driven approaches using optimal anomaly detection techniques for credit card fraud detection. The efficiency of this method is studied over a real data set from European credit card holders. Our experiments show that our approaches achieved a high-level of detection accuracy and a low percentage of false alarm rate. Our approaches will bring many benefits for the organizations and for individual users in terms of cost and time efficiency.

Anomaly detection in wireless sensor networks via support vector data description with mahalanobis kernels and discriminative adjustment

In the past few years, wireless sensor networks (WSNs) have been increasingly gaining impact in the real world with with various applications such as healthcare, condition monitoring, control networks, etc. Anomaly detection in WSNs is an important aspect of data analysis in order to identify data items which does not conform to an expected pattern or other items in a dataset. This paper describes a anomaly detection method using support vector data description (SVDD) kernelized by Mahalanobis distance with adjusted discriminant threshold. The efficiency of this method is studied over a real data set. Numerical result demonstrates that the proposed approach achieved a high-level of detection accuracy and a low percentage of false alarm rate owing to wise choices of discriminant threshold.