First course of its kind prepares students for big data in manufacturing
2/1/2019
Manufacturing collects the largest dataset of any industry. That data, often high-dimensional or noisy, goes underutilized because of a lack of adequate tools to analyze it. As professionals entering the manufacturing industry, many graduates are poorly prepared for tackling the big data analytics that are becoming more and more prevalent in manufacturing.
All the work done in the course utilizes data taken from real-world manufacturing situations—including a Caterpillar plant in China; the nanomanufacturing (nanoMFG) node; MechSE colleagues’ research; and Shao’s own research. These datasets are used in lecture, homework, and projects. Shao continuously revisits each problem throughout the semester, but he utilizes different approaches each time. This gives the students a comprehensive view of the problem and an understanding of the strengths and weaknesses of each data processing strategy.
“I always give them many examples from my research. I always tell them, look I have this problem – what do you think?” Shao said. “I always engage them in the examples, the problems.”
“I try to teach my students how to be ready—not only on a technical knowledge level, but also in professional experience,” Shao said. “How do you communicate? In industry you don’t get credit if you don’t present it well. That’s very important.”
The course teaches students about statistical quality control and machine learning methods. In a single semester students cannot become experts in every data processing philosophy, but Shao said they will be acquainted with enough strategies that when they encounter a problem in the workforce they will know where to start.