 |
|
(The following description was used when this course was taught during the 2007 Semester. There will likely be differences in the instructor, textbook, and/or outline the next time the course is taught.)
Instructor: Dr. Yu Ding ,
Reference Books
On multivariate monitoring: R. A. Johnson, R. A., and D. W. Wichern (2001). Applied Multivariate Statistical Analysis (5 th Edition) , Upper Saddle River , NJ : Prentice Hall.
On statistical process control : D. C. Montgomery (2003). Introduction to Statistical Quality Control (5 rd Edition). New York City : John Wily and Sons.
On scan statistics : J. Glaz, J. Naus, and S. Wallenstein (2001). Scan Statistics , New York City : Springer.
Useful software : MATLAB and R (freeware).
Description: Fundamental methods about anomaly and change detection in a process or an environment. Methods covered include the univariate and multivariate analysis for continuous and discrete data, the data pre-analysis methods (such as dimension reduction), and the scan statistics. Methods of anomaly and change detection find themselves in a broad spectrum of applications, including manufacturing quality control, health care delivery, as well as homeland security.
Course Topics
- Basic mathematical setup
- Univariate analysis: Shewhart chart, CUSUM and EWMA
- Univariate analysis: Generalized likelihood ratio
- Univariate analysis: Risk adjustment
- Multivariate analysis: T2 , multivariate EWMA and CUSUM
- Multivariate analysis: Data reduction and projection
- Multivariate analysis: Handling the profile signals
- Time-sequence analysis: Scan statistics
- Handling discrete data
- Applications (QC, healthcare, security applications)
Preferred Background
- Linear Algebra or Matrix Algebra
- Knowledge on hypothesis test, likelihood function, and linear regression (ISEN 414/616, STAT 608, or equivalent).
