Analyzing the Internet of Things
Taking advantage of the IoT's enormous data stream takes some sharp analytics.
This program was produced by the Marketing Department of WIRED and Ars in collaboration with CA Technologies.
The billions of data-collecting devices that comprise the Internet of Things (IoT) have big implications for business. From factories to farms, using everything from environmental sensors to remote asset tracking beacons, the IoT gathers critical information that can help many companies improve their processes quickly and efficiently.
Businesses have quickly learned, however, that the velocity of the incoming data streams can be overwhelming. It’s not enough to simply gather up an ocean of data; it needs to be smartly analyzed to make intelligent decisions.
That’s where predictive analytics comes in. Predictive analytics, simply put, uses algorithms and complex modeling to identify future outcomes based on historical data, and experts say it should be an essential part of corporate strategy. “Every business decision that’s made—every one—should try to have predictive analytics at the center of it,” said Jeff Ma, director of business insights at Twitter, and a predictive analytics expert, whose work has been documented in books and movies.
Every business decision that’s made—every one—should try to have predictive analytics at the center of it.
— Jeff Ma, Director of Business Insights, Twitter
The smart use of predictive analytics can accelerate the opportunity provided by the IoT by turning raw data into actionable knowledge, noted one analyst at a major research firm. “Analytics will play a big role in deriving value from the steady stream of IoT data,” the research director at large analytics firm said. “The value lies not so much in the data itself as in what’s done with it.”
Predictive analytics can help make an informed and dispassionate use of data, helping avoid the cognitive biases that can sometimes cloud sound business decisions. These natural human biases include things such as:
- The omission bias (the tendency to not do something for fear it will make a poor situation worse),
- The illusion of control (overestimating one’s influence over external events),
- The hindsight bias (the tendency to see past events as predictable), and
- The normalcy bias (a refusal to plan for something that has never happened before).
Any of these, and dozens of others, can harm or even destroy a company.
Instead, the larger question should be: “Now that we have more and more devices and machines supplying us with data, we need to ask ourselves what great things we can do with it,” Ma said.
Examples of IoT-based analytics include the sensors used by large retailers that push coupons to shoppers via their mobile apps as they near particular items in the store. IoT devices may also pay a key part in predictive maintenance; as data is streamed from multiple sensors it is fed into computer models that tells maintenance technicians when, for instance, a piece of farm or factory equipment might fail. Automakers, too, are equipping cars that alert customers when tire pressure is low.
Analytics of the data stream can happen at multiple points. Typically, thousands of data-producing sensors feed their information to a gateway where it is then relayed to the cloud. Data reduction and analytics could happen at any of a number of points along the information stream—from data-gathering point to cloud gateway to an aggregated database back at headquarters—depending upon how the analytics models are created.
Security and privacy concerns
Despite the potential for game-changing business insights, there are many concerns are security and privacy concerns with IoT. Since the IoT is, essentially, a “system of systems” it can be more vulnerable than a single managed system.
In recent months, researchers have demonstrated that, for instance, Internet connected cars can be hacked, even while in motion. Autos can be unlocked and even shut down while in motion. Last year, too, there were reports of serious vulnerabilities in Wi-Fi-dependent baby monitors that allowed hackers to monitor live feeds of kids sleeping in cribs. There are also fears that wearable devices could present a threat to privacy.
Experts agree that they need to maintain the confidentiality and integrity of both business and personal data collected within the IoT through strong encryption, authentication and data integrity protections. The FBI has also issued guidelines to increase the safety of those who use IoT-generated data.
That rapidly increasing flow of outside data into business—by 2020, a major research firm (Gartner) expects more than 26 billion units will make up the IoT—could dramatically reshape information flow and organizational structure. The increasing use of IPv6 and the expanding deployment of Wi-Fi networks will provide the accelerants, and the need to understand and move on the data flow will be critical.
Some foresee a time when a Chief Analytics Officer may be equivalent of a Chief Executive Officer. “Analytics is already a lot of what a CEO does,” Ma said. “A CEO is processing a lot of information and making their best decisions based on that. Many organizations already have a Chief Analytics Officer. This could be a real leadership role. We’re not quite there yet, but I think we’re heading in that direction.”