This fall, American University’s Department of Mathematics and Statistics hosted the prestigious XIV International Workshop on Intelligent Statistical Quality Control (ISQC). 

Scientists from top universities in the United States, Canada, Mexico, Japan, Germany, Italy, Portugal, and Spain attended the conference and presented innovative research results and sophisticated statistical methods, combining advanced knowledge of mathematics, statistics, and data science to tackle modern challenges of quality control. 

Statistical Quality Control Problems and Solutions

Statistical quality control refers to all types of statistical methods directed at monitoring and maintaining the high quality of production. “Monitoring the process involves collection and statistical analysis of massive data in real time,” says Mathematics and Statistics Department Chair and Professor Michael Baron. “When any parameters of the observed process jump out of the normal range or indicate peculiar behavior, it is important to know it as soon as possible.”

When collected data suddenly signal unusual patterns, the situation calls for quick decisions and urgent actions. Unexpected trends may indicate an emergency that is already happening or coming soon. Intelligent statistical quality control methods are applied far beyond the problems of industrial manufacturing. Wouldn’t we like to be alerted to an upcoming epidemic as soon as possible? Or social unrest? Exceptional rates of water pollution? Unusually fast patterns of ice melting? 

Participating scientists showed cutting-edge applications of statistical monitoring methods in many fields, from civil engineering, automotive industry, and market risk analysis to global climate, public health surveillance, and artificial intelligence. 

Theoretical Sophistication and Elegant Derivations 

Few people realize the depth of mathematical science behind these practical statistical solutions. Results and techniques from functional analysis, non-Euclidean geometry, theory of manifolds, and other areas of mathematics were effectively used by the conference participants in their elegant derivations of statistical monitoring tools.

Why does one need such a high level of theoretical sophistication for practical problems of quality control? Well, sometimes the new patterns are seen clearly, as on the graph on the left. But sometimes disruptions cannot be noticed by naked eye, as on the graph on the right. That is because the data are usually random and noisy. However, even very small disturbances can produce major damage if the new pattern remains. Thus, even slight changes are important to detect, which requires sensitive statistical techniques, on one hand, whichwould not produce frequent false alarms, on the other hand. Add complicated multi-dimensional data structures on top of this task, and you will face a problem that can only be solved by combining classical theory with modern methodology and computing power. 

Long and Rich History across Countries and Continents 

Taking a well-deserved break from their theories and methodologies, the conference guests explored famous tourist attractions of Washington, DC, in a 4.5-hour excursion with experienced tour guide Maricar Donato. 

International Workshops on Intelligent Statistical Quality Control have a long and rich history. Since 1980, they have taken place every three years in different countries and on different continents. ISQC workshops have been held in Germany, Denmark, Japan, Canada, Poland, China, and Australia. Hong Kong welcomed the previous ISQC in 2019, which makes American University the first post-pandemic ISQC host. The next, ISQC XV, is scheduled in Lisbon, Portugal.