Data Science: Making Sense of Digital Manufacturing

It’s hard enough to understand any one thing, let alone the complex interactions between many things—each of which is constantly changing, through the other things acting on it.

That’s the promise of data science, a mashup of computing, mathematics, statistical analysis, and good old-fashioned critical thinking. Its knack is to detect signals, structure, and patterns from data sets to advance learning and decision making in every field—from the study of the universe and human genome through medicine, engineering, the stock market… and manufacturing.

Data science is the latest iteration of technique in humankind’s quest to get to the bottom of things. As such, it has emerged as one of the “Big Tools” of our 21st century world, alongside the Large Hadron Collider in physics and the Hubble Space Telescope in astronomy—tools that let us peer deep inside, or far away, to peel away the layers of the innermost workings of things.

Manufacturing has always been fiendishly complex, making it easy to throw up your hands in frustration at attempts to uncover root causes. Now, digitization has opened a window into its innumerable variables—by capturing the many data points that constitute any manufacturing process, from raw material provenance through production steps, work in process, yield rates, quality conformance, equipment effectiveness, and so forth, all the way through supply-chain planning and logistics.

That’s where data science steps in, with its intimidating kit of tools: linear regression, density estimation, clustering, decision trees, nearest neighbors, scoring engine, and so on. You’ll most likely not need to know much about them except this: these are the techniques that can decode the streaming, cellular composition of manufacturing, analyze connected elements and occurrences to make sense of what has happened—and is happening, and offer informed predictions of the probability of what is to come.

Data scientists are now integral parts of manufacturing teams. Glassdoor ranked the job number-one on its list of the 50 best jobs in America for 2018. The description of a data science course offered at UC Berkeley gives you a hint of what they know: “… will explore the data science lifecycle, including question formulation, data collection and cleaning, exploratory data analysis and visualization, statistical inference and prediction, and decision-making.” The class focus includes “… languages for transforming, querying and analyzing data; algorithms for machine learning…; principles behind creating informative data visualizations; statistical concepts of measurement error and prediction…”

The applications of data science to digital manufacturing are limitless. Real-time understanding of operational processes enables higher throughput and rapid pivots. Analyzing shop-floor sensor data lets manufacturers make compensating tweaks when quality veers from tolerance. Customer sentiment can be graphed to customized design and production. Supply-chain forecasts and delivery decisions can be optimized down to the individual stockkeeping unit and location level.

Data science will power today’s industrial renaissance. It is the “petri dish” of the experience economy, fostering in its culture a direct, flexible path from consumer want to manufactured fulfillment.

John Martin

John Martin writes about technology, business, science, and general-interest topics. A former U.S. correspondent for The Economist (Science & Technology), he writes for the private sector, universities, and media, and can be reached at jm@jmagency.com.

Latest posts by John Martin (see all)