Q&A: Analyzing Sensor Data for Informed, Real-Time Decisions

Sensor data is increasingly important in business intelligence. What's behind the rise in interest in this data source, how is sensor data different from big data, and what's ahead? TempoIQ's Andrew Cronk shares his insights and predictions.

Sensor data is increasing in importance for many enterprises, especially with the rising tide of the Internet of things. Andrew Cronk, CEO of TempoIQ, explains the difference between sensor data and big data, what tools are needed to analyze sensor data, who should manage it, and what's ahead.

BI This Week: Why are sensors a big deal? What types of businesses are using sensors and for what purpose?

Andrew Cronk: The megatrend underlying both the need and demand for sensors is something I call the Age of Accountability. In short, the Age of Accountability is about using data to create feedback loops, which means that data then informs decisions.

Ultimately, in the Age of Accountability, everything that can be measured will be measured. We are making huge progress when it comes to measurement. Sensors are inexpensive to purchase and connect, so many companies are capturing data first and figuring out how it might inform decisions later. Now that we are starting to capture all this data, it's time to use it to make decisions.

This need for sensor data insight is how TempoIQ quickly established a following in the renewable energy industry. Renewable energy typically requires a high upfront investment that is recouped (and profitable) over years, so it is critical that geothermal and solar companies, for instance, be able to measure and quantify savings -- and sensor data provides that raw data.

Sensors are important for the renewable energy businesses. Another area where sensors can really be instrumental is in medical devices. This is beyond the first wave of wearables like Jawbone and Fitbit, and into products that actually save lives such as glucose meters and wearable defibrillators. Using real-time sensor analytics, companies using these products can spot vital signs outside the norm and act immediately to save someone's life.

What factors are enabling this latest burst of sensor-enabled products?

It's not that the demand has shifted, rather that our ability to utilize sensors has changed. To use history as an example, if you look at YouTube, it wasn't the first website created to host and view videos. Rather, it came along at the right time – when high-speed Internet was ubiquitous and digital cameras (phones and flip cameras) were prevalent. In other words, it came along at a time when it was easy to both capture and view videos, and when the enabling infrastructure existed for its success.

We are seeing the same now in the world of sensors. As I mentioned, sensors are cheap to buy and deploy. Simultaneously, much of the world is covered in wireless. By deploying sensors at scale and connecting them efficiently, the economics are finally in place for massive adoption.

How is sensor data different from big data?

In general, the term big data is kind of an all-encompassing phrase to detail a large collection of data points. However, big data more commonly deals with information around people and places -- for example, what websites someone views. Marketers use it to better cater products to your search history or preferences. Because this type of big data refers to human-driven actions -- a fixed number of people visit a site in a month -- this type of data, although large, has finite limits.

Sensor data, on the other hand, is machine-driven and nearly infinite in terms of what you can measure. In solar production, for example, sensors measure everything from the radiance of the sun to line-by-line output on solar panels -- all of which generate data at a much higher rate and from a lot more sources.

What unique analysis tools are required to analyze sensor data?

One of the goals of sensor analytics is being able to diagnose problems and make decisions quicker. This leads to real-time monitoring and analytics, where a system needs to compare against performance thresholds and historical data as new information streams in.

To compare with big data, the challenging questions are around large-scale consumer actions and macro trends -- things that get analyzed on the scale of hours, days, and weeks. With sensor data, the main differentiator is that you want the data and analysis in real time -- a decision that results in substantially more predictions.

How are companies using sensor data to inform decisions in real time?

Most experience with sensors is in personal technology -- with Jawbone or a wireless scale that also measures their body mass index (BMI). These charts show progress over time -- allowing individuals to add a walk to their routine to meet fitness targets.

However, sensor data in the business world is used to track performance and inform decisions in real time. In healthcare, products such as the LifeVest wearable defibrillator add options to a patient's healthcare decisions. This device and other sensors in the healthcare field measure certain conditions of the body and alert patients if they are in danger of a sudden cardiac event.

Potentially, these devices will continue to aggregate patient information, finding new ways to detect an event and alert doctors and emergency personnel before it occurs.

In other businesses, sensors help allocate resources in real-time. If a solar panel is not performing within the normal range, sensors flag the underperforming panel and enable a decision: to do nothing, monitor it, or send a tech to fix it. Without proper analytics, techs could be dispatched at the first sign -- whether or not it is ultimately needed.

Within an organization, who should take ownership of real-time sensor data and maintenance?

Sensor data falls squarely within the CIO's role. Even though you historically think about the CTO when you think about data, the CIO is responsible for bringing data outside the company into the company. All these environmental factors that impact business need to be captured -- and that is the role of the CIO.

Inside this decision-making process, CIOs have an incredibly hard job. First, they must decide on the hardware and which sensors to deploy. Then, the CIO needs to determine how to connect the sensors with the company's network. Is it through cell towers or wireless Internet? Finally, there's the question of how to store, monitor, and analyze all this data.

Usually, responsibility for the data turns over to the CTO when it comes to how the company wants to show the data to the customer -- the dashboard and interface.

What's next? How big is this going to be?

This is the billion-dollar question, but it is likely that the market for sensors and sensor analytics is going to be larger than anyone currently believes. There is a lot of excitement surrounding retail sensors, but the benefits (besides the "cool" factor) have not been proven. In the business world, there is much more opportunity to revolutionize business models.

Hardware companies are not valued well in today's market; yet, if you look at the way that analytics has the potential to change "simple" hardware companies into hardware and software/analytics businesses, it can add substantially to the bottom line.

TDWI Membership

Get immediate access to training discounts, video library, BI Teams, Skills, Budget Report, and more

Individual, Student, & Team memberships available.