Robust control charts in industrial production of olive oil

Grilo, Luís M.; Mateus, Dina M. R.; Alves, Ana Catarina; Grilo, Helena L.

doi:10.1063/1.4897794

Cited by 3 publications

(4 citation statements)

References 6 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Less than -3 sigma 6 2 points out of the last 3 below -2 sigma 6 4 points out of the last 5 below -1 sigma 8 2 points out of the last 3 above +2 sigma 28 8 consecutive points above the centre line 29 4 points out of the last 5 above +1 sigma 29 8 consecutive points above the centre line 30 4 points out of the last 5 above +1 sigma 30 8 consecutive points above the centre line 7 points violate control rules.…”

Section: Rule Violations For Run (A) Case Numbermentioning

confidence: 99%

“…Violations for Points 5 4 points out of the last 5 below -1 sigma 6 4 points out of the last 5 below -1 sigma 7…”

Section: Rule Violations For Mr (A) Case Numbermentioning

confidence: 99%

“…However, the empirical distribution of PIwas not normal, which is an assumption commonly recommended in the literature [1,3,4,5,8], once it can cause serious errors in the calculation of I charts triggering false alarms (Type I errors) or misses (Type II errors).An adequate Box-Cox transformation (BCT) for normalization [2] is considered in some simulation studies that prove the efficiency and robustness of statistical estimators drivingto more robust control charts, in what regards the false alarm rate and the time required to detect the occurred changes [1,4,5,10,11].According to other studies the estimated control limits, under non-normality, may not be a serious problem (especially when we have light tails, moderate outliers and asymmetry), which made these type of Shewhart charts a robust tool [6,7,12,13,14,15].Despite the controversy of this assumption, we decided to obtain and compare I and MR charts for PI, before and after an adequate BCT, and also computing the limits based on average and on median of MR [12].…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Individual and moving range control charts in the production of olive oil

Grilo¹,

Grilo²

2015

AIP Conference Proceedings

Self Cite

View full text Add to dashboard Cite

Quality control charts are usually used to monitor peroxideindex of olive oil, which is a variableanalysedin order toassess the quality and tocharacterize the typeof oil. However, the empirical distribution of peroxideindex does not often verify the assumption of normality. In this case study, we obtained the individual and moving range charts for peroxideindex of virgin olive oil, computing the control limits using theaverage and also the median moving range. Afterwards, we compared these charts with the more efficient and robust ones, based on an adequate Box-Cox transformation. The conclusions obtained with these approaches are relatively different.

show abstract

Section: Rule Violations For Run (A) Case Numbermentioning

confidence: 99%

“…Violations for Points 5 4 points out of the last 5 below -1 sigma 6 4 points out of the last 5 below -1 sigma 7…”

Section: Rule Violations For Mr (A) Case Numbermentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Individual and moving range control charts in the production of olive oil

Grilo¹,

Grilo²

2015

AIP Conference Proceedings

Self Cite

View full text Add to dashboard Cite

show abstract

“…Fernanda Figueiredo [3] makes a comparison of sampling plans by variables using the bootstrap and Monte Carlo simulations. Luís M. Grilo [4]apply the bootstrap methodology to construct control charts, before and after a Box-Cox transformation, and compare their robustness and performance.João P. Martins [7] analyze the performance of a maximum likelihood computational algorithm for the prevalence rate estimation. Dora Prata Gomes [9] explores the statistical software R for modeling spatial extreme precipitation data.…”

Section: Faculdade De Ciências E Tecnologia and Cma -Universidade Nova mentioning

confidence: 99%

Preface of the "Symposium on computational statistical methods"

Caeiro¹

2014

AIP Conference Proceedings

View full text Add to dashboard Cite

In the last decades, computers power has increased exponentially and that power hasallowed the arisen of new and more complex computational statistical models and methods.In this symposium, we bring together researchers working or applying computational statistical. All the submitted papers in the symposium have been peer reviewed and selected on the basis of the originality, signi cance and clarity.In our symposium (by presenter order), Miguel Felgueiras [2] investigate Gaussian mixtures with independent components and introduce a shifted and scaled t-Student distribution as an approximation for the distribution of Gaussian mixtures. Filipe J. Marques [6] study the distribution of the linear combination ofindependent Gamma random variables. Ayana Mateus [8] presents an R package implementation of several nonparametric randomness tests. Fernanda Figueiredo [3] makes a comparison of sampling plans by variables using the bootstrap and Monte Carlo simulations. Luís M. Grilo [4]apply the bootstrap methodology to construct control charts, before and after a Box-Cox transformation, and compare their robustness and performance.João P. Martins [7] analyze the performance of a maximum likelihood computational algorithm for the prevalence rate estimation. Dora Prata Gomes [9] explores the statistical software R for modeling spatial extreme precipitation data. Finally, Frederico Caeiro [1]employ the Monte Carlo methodology to compare the finite sample behavior of two reduced bias estimators of the extreme value index.

show abstract