What's the difference between analysis and analytics? One looks at the past while the other tries to predict the future.
The distinction between analysis and analytics is often blurred and more often misunderstood. Indeed, engineers may think that analytics is for marketers, while engineers do analysis. While there's some truth to that, engineers can perform analytics that can help with manufacturing and reliability.
With the continuing shift toward the collection of massive amounts of data and more powerful tools to extract hidden insights, it can be worthwhile to revisit the definitive and separate contributions of "analysis" versus "analytics." As we look ahead toward new advanced analytical capabilities, such as predictive and prescriptive analytics, solidifying these fundamental terms can be a good starting point in understanding what is possible.
Data analysis: What happened?
According to the Merriam-Webster dictionary, analysis is "a detailed examination of anything complex in order to understand its nature or to determine its essential features: a thorough study." Analytics is defined as "the method of logical analysis." Embedded in these definitions is an inherent aspect of past and future assessment. For a "detailed examination" or a "thorough study" to take place, the data must exist and occur in the past. The question that data analysis answers, whether based on a single data set or thousands of data sets, is, "What happened?" For example, an automotive brand uses analysis — after the fact — to determine which cars, models, and geographic locations required a recall. Analysis can be considered an in-depth review and sorting of the current facts, which may be a more-than-sufficient assessment for many decision-making scenarios.
Measurement data of LED failures can show how increased pulse power causes shorter time to failure.
Data analytics: Why did it happen and what will happen next?
Merriam-Webster’s definition of analytics as a "method of logical analysis" includes the term analysis, but introduces a significant differentiator with the term "logical." Analytic methods use data to answer questions that occurred in the past, but also provide insights or deductive reasoning to act in the future. Gartner Research covers analytics in its industry reports and defines the word in its IT Glossary. As Gartner notes, "Increasingly, analytics is used to describe statistical and mathematical data analysis that clusters, segments, scores, and predicts what scenarios are most likely to happen."
In the automotive-brand scenario, analytics users would employ advanced machine-learning algorithms to compare and correlate a broad range of data for recalled cars versus non-recalled cars, such as date of manufacture, environmental influences, and components used. With in-depth analytics, the manufacturers will be able to go beyond a historical review, anticipate or predict future scenarios that would result in a recall, and perhaps take the necessary steps to minimize the negative impact of a recall.
Using data analytics from manufacturing can help predict which parts might fail in the field, causing RMAs.
Advanced analytics: How do we fix it?
Big data aggregation and sophisticated machine-learning analytics tools provide the capability for organizations to reveal previously unknown patterns, correlations, and other "hidden" logistical information. These capabilities are also a powerful driver for the next phase of analytics: predictive and prescriptive.
Real-time advanced analytics empowered by data sourced throughout a global supply chain enables the predictive ability to pose and answer the question, "What will happen next?" With this valuable predictive knowledge comes the ability to be prescriptive and make permanent system-wide changes that will prevent costly recurrences of inefficiencies or errors. For manufacturers, predictive and prescriptive analytics anticipates problems and presents tangible solutions for correcting root causes of product issues, which has a significant impact on production yield and product quality.
What about individual product quality?
Analytics are now at a juncture where the questions being asked were not previously considered because there wasn’t the prospect of finding an answer. Big-data analytics are now able to examine enormous data sets, detect hidden patterns, and identify "needle-in-the-haystack" correlations that can provide an unprecedented knowledge base that gives answers to previously unanswerable questions.
One of the new questions that can be answered with big data analytics is, "Is my product quality really good?" Much of the analytical focus to date has been on process improvement, but real-time product analytics is proving to be a critical and useful complement to the traditional process-improvement analytics. A good example is in semiconductor manufacturing, where product analytics can capture the full genealogy of every device in generating predictive and prescriptive analytics to ensure minimal defects. In addition, by using product-centric data, root cause analysis and traceability is possible not only for the individual device but also for all of the downstream electronics devices that utilize those semiconductor components.
Big-data solutions, sophisticated data governance, and advanced analytics tools are driving operational intelligence to deeper and unprecedented levels. Analysis and analytics will both have a major influence in the realm of Industry 4.0.