All Things Considered: The Analytics of Things

The Internet of Things will enable your organization to capture vast troves of data. How will you make sense of it all?

The Analytics of Things

Metcalfe’s Law describes the value of networks. Two nodes can make one connection, delivering some value. Five nodes can achieve 10 connections, delivering more value. One thousand nodes can yield 499,500 connections, delivering even greater value. And so on.

That’s true for phones in a telco network, and it’s roughly true for sensors in an Internet of Things (IoT) network, depending on how interconnected the sensors are. But what about the data those sensors generate? Do 1 million data points provide value? Do 100 million data points deliver 100 times more value?

It all depends on your ability to analyze the data. And that has people looking beyond IoT to the analysis of IoT data. Gartner refers to this as “the Information of Everything.” Analytics guru Tom Davenport prefers “the Analytics of Things.”

“The Internet of Things is about getting data from one place to another,” says Davenport, who has been named among the Top 100 most influential people in IT by Ziff Davis. “The Analytics of Things is about making sense of that data and taking action on it.”

A Fine Mesh

device mesh
The onslaught of data begins with what Gartner refers to as the “device mesh.” The device mesh “refers to an expanding set of endpoints people use to access applications and information, or interact with people, social communities, governments and businesses.”1 Beyond desktops and mobile devices, the device mesh will include wearables, home electronics and appliances, transportation devices and “environmental” devices such as cameras.

“As the device mesh evolves, we expect connection models to expand and greater cooperative interaction between devices to emerge,” Gartner reports. “We expect significant innovation in new types of devices during the next five years. This will create many new digital business opportunities, but also pose significant IT security and management challenges.”2

The device mesh leads directly to Gartner’s Information of Everything. “Harnessing this information requires organizations to understand the flood of unstructured information that goes beyond textual, audio and video information to include sensory and contextual information,” Gartner says.3 In other words, data analytics will need to become much more sophisticated and automated.

Industry Analytics

The Analytics of Things isn’t just for industries such as manufacturing that are becoming heavy users of IoT sensors. It will cut across categories, from retail to financial services to public sector.

“Today, retail uses RFID tags,” notes Davenport, who is the President’s Distinguished Professor of Information Technology and Management at Babson College. “In the future, they’ll become smart tags and will be connected to the Internet. Every industry will use smart tags.”

Whether they capture data in central repositories or process it in sensors at “the edge,” all organizations will begin to collect more data than they can manage using old paradigms. “Today, most organizations are doing only simple analyses, with SQL-like rules,” Davenport says. “As they gather more data from more sources, they’ll need more sophisticated approaches, like deep learning.”

The Internet of Things is about getting data from one place to another. The Analytics of Things is about making sense of that data and taking action on it.
Deep learning is a branch of artificial intelligence that combines specialized software with massive compute power to “learn” from the experience of analyzing data to recognize patterns in natural language, images and other unstructured data. The DeepMind division of Google®, which the company acquired in 2014 for a reported £400 million, uses deep learning in its AlphaGo system, which famously defeated South Korean “Go” champion Lee Sedol in March 2016.

Connecting the Bits

What steps should you take to start achieving the Analytics of Things? The first is data integration, and this is where many companies have trouble, Davenport says. IoT devices produce data in many formats — the average car has well over 100 sensors, and nearly as many data formats. Unless you plan to analyze each stream separately, you need to integrate the data.

Next, your analytics need to do more than simply report a condition. They also need to tell you the underlying causes of the condition. Case in point: “Companies are finding that IoT sensors aren’t always reliable,” notes Davenport. This is true, for example, in the healthcare industry, where false alerts create a condition known as alert fatigue. “So increasingly, companies will need analytics to determine whether alerts should be acted on,” Davenport says.

Finally, you need to connect your IoT data with your business data. Let’s say you’re a consumer products company collecting data from Internet-attached appliances. You have to link sensor output with who owns the appliance, where the nearest repair depot is, which maintenance agreement is in place, and so on.

In the end, though, the Analytics of Things isn’t strictly about analytics any more than the Internet of Things is about the Internet. It’s ultimately about identifying and capitalizing on new business models, or understanding and serving customers more effectively. It’s these imperatives that will drive companies to embrace both the Internet and the Analytics of Things.

1, 2, 3 “Top 10 Strategic Technology Trends for 2016: At a Glance,” Gartner, October 2015
4, 5, 6 “Worldwide Internet of Things Forecast, 2015 – 2020,” IDC, July 2015
7 “Spoiler Alert! Everyone Needs an IoT Business Plan,” IDC, March 2015
8 “IDC Reveals Worldwide IoT Predictions for 2015,” IDC, January 2015
Research and graphics used in this story are the property of their owners and are used with permission.

Third parties quoted in this article are quoted with permission.

DeepMindTM is a trademark and Google® is a registered trademark of Google, Inc.
Gartner® is a registered trademark of Gartner, Inc.
IDC® is a registered trademark of International Data Group, Inc.
Ziff Davis® is a registered trademark of Ziff Davis LLC.

Featured

[x] Close

Sign Up for Email