### Abstract

Monitoring real-time data steams is an important learning task in numerous disciplines. Traditional process-monitoring techniques are challenged by increasingly complex, high-dimensional data, mixed categorical and numerical variables, non-Gaussian distributions, nonlinear relationships, etc. A new monitoring method based on real-time contrasts (RTC) between the reference and real-time data is presented. RTC assigns one class label to the reference data, and another class label to a window of real-time data, and, thus, transforms the monitoring problem to a dynamic series of classification problems. This differs from previous work that trained a classifier one time at the start of monitoring. Furthermore, a number of monitoring statistics based on the generalized likelihood-ratio principle are discussed. These include error rates and class probability estimates from classifiers. Typically, the window size for the real-time data is much smaller than the size of the reference data and this class imbalance is also considered in the approach. Variable contributor diagnostics can be simultaneously obtained from the approach and are briefly discussed. Both mean and variance shifts are illustrated. Experiments are used to compare monitoring statistics and to illustrate performance advantages of RTC relative to alternative methods.

Original language | English (US) |
---|---|

Pages (from-to) | 9-27 |

Number of pages | 19 |

Journal | Journal of Quality Technology |

Volume | 44 |

Issue number | 1 |

State | Published - Jan 2012 |

### Fingerprint

### Keywords

- Control charts
- Generalized likelihood ratio
- Multivariate control charts
- Nonparametric
- Statistical process control
- Supervised learning

### ASJC Scopus subject areas

- Industrial and Manufacturing Engineering
- Safety, Risk, Reliability and Quality
- Strategy and Management
- Management Science and Operations Research

### Cite this

*Journal of Quality Technology*,

*44*(1), 9-27.

**System monitoring with real-time contrasts.** / Deng, Houtao; Runger, George; Tuv, Eugene.

Research output: Contribution to journal › Article

*Journal of Quality Technology*, vol. 44, no. 1, pp. 9-27.

}

TY - JOUR

T1 - System monitoring with real-time contrasts

AU - Deng, Houtao

AU - Runger, George

AU - Tuv, Eugene

PY - 2012/1

Y1 - 2012/1

N2 - Monitoring real-time data steams is an important learning task in numerous disciplines. Traditional process-monitoring techniques are challenged by increasingly complex, high-dimensional data, mixed categorical and numerical variables, non-Gaussian distributions, nonlinear relationships, etc. A new monitoring method based on real-time contrasts (RTC) between the reference and real-time data is presented. RTC assigns one class label to the reference data, and another class label to a window of real-time data, and, thus, transforms the monitoring problem to a dynamic series of classification problems. This differs from previous work that trained a classifier one time at the start of monitoring. Furthermore, a number of monitoring statistics based on the generalized likelihood-ratio principle are discussed. These include error rates and class probability estimates from classifiers. Typically, the window size for the real-time data is much smaller than the size of the reference data and this class imbalance is also considered in the approach. Variable contributor diagnostics can be simultaneously obtained from the approach and are briefly discussed. Both mean and variance shifts are illustrated. Experiments are used to compare monitoring statistics and to illustrate performance advantages of RTC relative to alternative methods.

AB - Monitoring real-time data steams is an important learning task in numerous disciplines. Traditional process-monitoring techniques are challenged by increasingly complex, high-dimensional data, mixed categorical and numerical variables, non-Gaussian distributions, nonlinear relationships, etc. A new monitoring method based on real-time contrasts (RTC) between the reference and real-time data is presented. RTC assigns one class label to the reference data, and another class label to a window of real-time data, and, thus, transforms the monitoring problem to a dynamic series of classification problems. This differs from previous work that trained a classifier one time at the start of monitoring. Furthermore, a number of monitoring statistics based on the generalized likelihood-ratio principle are discussed. These include error rates and class probability estimates from classifiers. Typically, the window size for the real-time data is much smaller than the size of the reference data and this class imbalance is also considered in the approach. Variable contributor diagnostics can be simultaneously obtained from the approach and are briefly discussed. Both mean and variance shifts are illustrated. Experiments are used to compare monitoring statistics and to illustrate performance advantages of RTC relative to alternative methods.

KW - Control charts

KW - Generalized likelihood ratio

KW - Multivariate control charts

KW - Nonparametric

KW - Statistical process control

KW - Supervised learning

UR - http://www.scopus.com/inward/record.url?scp=84857529385&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84857529385&partnerID=8YFLogxK

M3 - Article

AN - SCOPUS:84857529385

VL - 44

SP - 9

EP - 27

JO - Journal of Quality Technology

JF - Journal of Quality Technology

SN - 0022-4065

IS - 1

ER -