Leveraging Big Data

Leveraging Big Data

When companies define specific business challenges they want to solve, then they're poised to take advantage of data analytics.

What Makes Big Data Big?

Data scientists refer to the elements of big data as “the three V’s”:

Volume. There’s a lot of it. An IBM estimate says that most businesses have more than 100,000 gigabytes of stored data. IBM also estimates that 2.2 billion gigabytes of data are created every day. Other sources assert that number is higher and climbing.

Velocity. This means that all that data is coming in really quickly.

Variety. And it’s coming from numerous sources: social media, sensors, emails, text messages, online transactions, and so on.

Founded in 1945, metal fabricator Dalsin Industries makes custom parts for several sectors, notably transportation, medical, audio-visual, solar, and energy generation. “We offer just about anything a customer can dream up in terms of small parts and large assemblies,” says Keith Diekmann, Dalsin’s vice president of technical operations.

Dalsin also knows how to innovate. In September, The Silicon Review, a magazine that covers business and high tech, listed the Bloomington–based manufacturer among its Top 50 Companies to Watch in 2019. The magazine cited Dalsin’s use of automation and robotics, as well as its deployment of a global supply chain program. One of Dalsin’s products is the Memphis Wood Fire Grill, which incorporates a mobile app that allows residential chefs to remotely monitor and adjust cooking temperatures.

Dalsin has “always been on that leading edge of using technology to improve processes and become more efficient and effective in making the parts,” Diekmann says. Its production operation incorporates robotic welding and 3D printing, which allows the company to reduce the lead time for its customers.

Now Dalsin, which employs about 200 people, is exploring what constitutes the next generation of manufacturing. Sensors and other electronic measurement devices are generating an abundance of information about the company’s production equipment. “A lot of our machines are connected, collecting that data,” Diekmann says. “We’re continually analyzing it, trying to figure out what this tells us, and how we can use this information in the future.”

Dalsin is just beginning to learn what it could do with this data. Businesses in many other industries are in the same boat. This is the realm of what’s been called “big data”: It includes not only data management and analytics but also more advanced technologies, notably the internet of things (IoT), machine learning, and artificial intelligence (AI). Those latter terms conjure up images of self-driving cars and sentient robots—including, in many people’s minds, the scary dystopia of The Terminator movies.

In the real world, business applications of big data technologies are more down to earth. Some companies simply are looking for a better way to store and access all of the data they’ve gathered. Many want to use it to better understand and sell to their customers or want to figure out how to make their processes more efficient. And many want to forecast who might buy a product or when a crucial machine might need maintenance.

Business uses of big data promise greater operational efficiency and customer insight. But the field of data analytics is still evolving, and companies need to determine if there’s a real business application before committing the time and money to data projects.

What’s the use?

Data scientists typically distinguish between two types of data:

  • Structured data is data that is formatted and organized, usually in a database. Think of transaction data, for instance.
  • Unstructured data can’t be organized in a traditional database. Examples include emails, texts, Facebook comments—data that could provide useful insights to a business, but which has to be “structured” using various nonstandard formats.

There’s a massive amount of both kinds of data to sift through. But there are several tools that can help make that data valuable to businesses. Cloud computing, for example, allows companies to store more data to help identify useful patterns. “Because so much computing power has become available to us, we’re able to solve a large number of [business] problems,” says Bhabani Misra, associate dean of graduate programs in the School of Engineering at the University of St. Thomas. The data engineers his school graduates can apply their skills in just about any industry, including health care, hospitality, manufacturing, retail, financial services, and agriculture.

Digital technology allows companies to gather more of both types of data than ever before. At the Nerdery, a Bloomington–based software development and technology consultancy, the lion’s share of its business clients are looking for ways to manage all their data.

This might involve what Nerdery data science director Justin Richie terms a “data lake.” That’s a data storage system, usually cloud-based, that allows companies to store their data without “connecting” it. The company pools this data in order to derive insights later, Richie says. For a retail client, that “lake” can be filled with orders, product SKU information, and information unrelated to ordering, such as data from HR or social media marketing, he notes.

Data management and analysis are just the beginning when it comes to business uses. The next step involves taking that information and programming software and hardware to “teach itself” to recognize patterns in and derive insights from the data. “The whole premise of machine learning is that we learn from past examples and outcomes,” says Ravi Bapna, a business analytics and information systems professor at the University of Minnesota’s Carlson School of Management. While using this data, “we can let the machine work really hard with the math and science to find the rules so that we can be ready for new situations that come up,” Bapna says.

Machine learning is a key attribute of AI. Bapna notes the hype around what’s called general AI—machines becoming as intelligent as or more intelligent than humans. “But that’s not where the sources of business value are,” Bapna says. He argues that the real value of big data insights comes from “narrow AI,” which applies data analysis and machine learning to very specific business tasks.

Historic data can help a manufacturer predict whether and when a production machine will fail. “If it’s a capital-intensive asset, I don’t want unplanned downtime,” Bapna says. “If I have a good algorithm that can predict that failure, then I can perform maintenance and save the company money.”

A related set of business uses falls under the term “internet of things.” Simply defined, IoT refers to the online monitoring and management of equipment and infrastructure. “When that equipment is integrated into workflow, that’s when you really get your first big data opportunity,” says Scott Nelson, chief product officer at Digi International, a Hopkins–based tech firm that specializes in IoT products and services. Perhaps the most common application, Nelson says, is using sensors and data to predict when maintenance might be required. When might my truck go down? Or even: When might my company’s break-room ice machine stop producing ice and need repair? “That’s where AI starts to come in,” he says.

Case in point: oil and gas infrastructure. When prices are down, these companies are less likely to invest in new equipment and add more employees, and instead seek to optimize what they have. “They’ll move into a big data scenario when they have enough deployments that they can start looking at different levels of performance in different locations,” Nelson notes. Data analysis and insights can lower the risk of adverse consequences of any operational decision, he adds.

Big data insights also can help businesses compete more effectively. St. Louis Park–based custom software development company Coherent Solutions has a client that’s an online company that delivers restaurant food. It’s a growing industry with increasing competition, and to stay in the fight, the client needs to analyze its customers’ information to market to them successfully. The data the client gathers includes social media comments, order history, and customer demographics.

“All of that needs to be processed reliably and quickly” to offer restaurant and food suggestions to customers via the food company’s app, says Lenny Kotlyar, Coherent Solutions’ director of project delivery methodology. “For that client, we built a platform on the cloud to load and ‘clean’ the information so that it can be used to provide insights—and give suggestions to their customers.”

Defining data targets

For company managers who want to learn more about big data and its potential business uses, there are several short courses that can get them up to speed, without requiring them to become data scientists themselves. The Carlson School, for instance, offers Business Analytics for Leaders, a three-day executive education course whose content can be customized for different companies. Those businesses have included 3M, Optum, and Allianz.

“Often what we’re finding, at least with the medium- to large-size companies, is that they’re better off getting a group of people together who understand these concepts—perhaps from different functions, such as IT, HR, other aspects of the business,” Bapna says. “Then it’s much faster to adopt.” He contrasts that scenario to an executive who completes a program or workshop on her own and then has to report on what she’s learned to the rest of the company.

The Opus College of Business at the University of St. Thomas offers a nondegree business analytics and data visualization course though its executive education program. It’s designed to help executives understand the importance of data and how data insights can solve common business challenges.

As a tech consultancy, the Nerdery also advises clients interested in learning more about how they can use their data as the foundation for machine learning and AI applications. One of the Nerdery’s roles is helping these businesses “cut through the buzzwords and give them a more tangible-use case,” Richie says.

Start by identifying a real business problem. “If you don’t have a laser focus on what you’re doing with your data, it can be very easy to get distracted, and you will fail in that initiative,” he says. In 2016, Gartner, a Connecticut–based technology research firm, reported that 60 percent of big data analytics initiatives fail. “I would associate that [failure rate] largely with the strategy that they started with: They were pursuing these kinds of technologies for technology’s sake,” Richie says.

That observation segues into Richie’s second recommendation: Start small and prove value. “Data analytics is as much a cultural implementation as a technological one,” he says. “Helping organizations be data-driven is almost as critical as doing the technology framework underneath it.” Most customers the Nerdery talks to are still managing their digital data primarily by manual means. “We can help them unlock that by automating,” Richie says.

Coherent Solutions’ Kotlyar says big data-related projects “need to be very carefully evaluated and analyzed, just like any other business case where technology is applied.” If there’s no real business benefit, “you’ll end up with something that is either inefficient from a cost standpoint, or you may end up with a cool technology that you can’t really apply.”

Digi’s Nelson views big data and the technologies related to it as “a rich environment for innovation, for small steps that lead to large gains.” Businesses don’t need to see big data and its various uses as disruptive transformation, he adds. A data-based technology platform like IoT can be the breeding ground of “incremental, evolutionary innovation” that can help companies improve their operations.

And that’s the kind of innovation companies like Dalsin are starting to explore.

Gene Rebeck is TCB’s northern Minnesota correspondent.