The Changepoint Model for Statistical Process Control

Hawkins, Douglas M.; Qiu, Peihua; Kang, Chang Wook

doi:10.1080/00224065.2003.11980233

Cited by 319 publications

(189 citation statements)

References 16 publications

Supporting

Mentioning

182

Contrasting

Unclassified

Order By: Relevance

“…a nite sequence of independent random variables (X , ·, Xn), where x t is a particular realization of X i at the point in time (t i ), the segments before {X , ·, X k } and after {X k+ , ·, Xn} a breakpoint (T = k) the distribution of the data is di erent, and this change can indeed be detected using a two sample hypothesis test [28]. In the present study, since the modelled time series were of Gaussian distribution, the Student-ttuned for detecting changes in the mean [36] -Bartlett -designed for nding changes in the variance [50] -and Generalized Likelihood Ratio (GLR) statistics -originally tuned for detecting changes in both [51] -were used and compared on the di erently modi ed time series. The Average Run Length (ARL0) -as described in quality control literature [53] -was set to 200, which corresponds to α=0.95, the startup sequence (S) was set to 2,8,10 (corresponding to the default setting 10%=20 pcs.…”

Section: Sequential Changepoint Detection Via the Cpm Methodsmentioning

confidence: 99%

“…The foundations of the Change Point Model (CPM) framework were laid by Hawkins et al [36], and Hawkins & Zamba [50] to detect changepoints in variance and/or the mean of Gaussian random variables. Numerous CPMs have so far been developed to be applied within this framework to suit the conditions of variously distributed data: parametric, with breakpoints of di erent origin such as shifts in mean or variance [51], or non-parametric [52], as well.…”

Section: Sequential Changepoint Detection Via the Cpm Methodsmentioning

confidence: 99%

“…The scope of the present study therefore lls a vacant niche in the topic by comparing the sensitivity of a kink point detection method based on trend analysis [7], a modi ed cross-entropy (CE) method [16], and an additional change point model framework (CMP), as introduced by Hawkins et al [36], and further developed by Ross et al [37]. It is supposed that due to their di erent computational schemes, the di erent models will be sensitive in di ering degrees to BPs of various origins.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Detecting breakpoints in artificially modified- and real-life time series using three state-of-the-art methods

et al. 2016

View full text Add to dashboard Cite

Time series often contain breakpoints of di erent origin, i.e. breakpoints, caused by (i) shifts in trend, (ii) other changes in trend and/or, (iii) changes in variance. In the present study, arti cially generated time series with white and red noise structures are analyzed using three recently developed breakpoint detection methods. The time series are modi ed so that the exact "locations" of the arti cial breakpoints are prescribed, making it possible to evaluate the methods exactly. Hence, the study provides a deeper insight into the behaviour of the three di erent breakpoint detection methods. Utilizing this experience can help solving breakpoint detection problems in reallife data sets, as is demonstrated with two examples taken from the elds of paleoclimate research and petrology.

show abstract

Section: Sequential Changepoint Detection Via the Cpm Methodsmentioning

confidence: 99%

Section: Sequential Changepoint Detection Via the Cpm Methodsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Detecting breakpoints in artificially modified- and real-life time series using three state-of-the-art methods

et al. 2016

View full text Add to dashboard Cite

show abstract

“…In our work we employ Student-t test statistic, as in Hawkins et al (2003), which is used in order to detect mean changes in a Gaussian sequence. And finally we use Bartlett test statistic defined in Snedecor and Cochran (1989) and further improved in Hawkins and Zamba (2005), which is aimed at variance changes detection.…”

Section: Gaussian Sequence Modelsmentioning

confidence: 99%

Structural Distress Index: Structural Break Analysis of the Czech and Polish Stock Markets

Princ¹

2016

EFAJ

View full text Add to dashboard Cite

Abstract:The estimation of multiple structural break models is usually associated with identification of spurious break points, which are identified by universal algorithms. This leads to overvaluation of structural distress in financial markets represented by data series. The paper is focused on an estimation of the new index, which incorporates results of Student, Bartlett, GLR, Mann-Whitney, Mood, Lepage, Kolmogorov-Smirnov and finally Cramer-von-Mises tests statistics together. The new measure is named Structural Distress Index and evaluates a probability of structural break occurrence based on estimations of proposed models. SDI values show that Czech and Polish stock markets went through more instable period in 1990s than at the beginning of the global financial crisis in 2007. SDI measure is straightforward and can be easily explained, the highest values of SDI can identify the most important break points of the research period, which starts in year 1993 and ends in year 2014. Universality of SDI offers its further extension and application to further research of financial markets.

show abstract

“…We are concerned with ongoing monitoring to detect assignable causes in the process in phase II controlling. Useful recognitions of phase I and phase II applications have been studied already, for example, by Kang and Albin (2000), Woodall (2000), Hawkins et al (2003), Woodall et al (2004), Montgomery (2005), and Jensen et al (2006).…”

Section: Introductionmentioning

confidence: 99%