Knowledge is power: Why IoT is all about the consequences of data

big data
Image credit: chombosan /

Like knowledge, big data only makes a difference when it is applied. The same is true for IoT, which is not about the things but the data and the consequence of that data.

Knowledge is power. There’s an observation that is both simple and concerning. In physics lessons, we used to learn about kinetic and potential energy. Potential was the stuff that was accumulated waiting to be put to work, while kinetic was the result (often explosive but always compelling to teenage science geeks) of the application of that potential.

So, it is with knowledge – unapplied, it has little if any utility. Applied correctly with precision, timing and consideration, it can literally change the world.

In trawling social media, the work of Kirk D Borne over at Booz Allen passed by and it caused one of those moments of clarity which come by infrequently. In this work, Kirk (@KirkDBorne, Principal Data Scientist at Booz Allen) lays out five stages of data analytics. Why does this topic become relevant to the wide and wonderful world of the Internet of Things, you may ask? It comes down to the reality that IoT is all about the data and the consequence of that data.

In Dr Borne’s work, he proposes five stages of analytics as an approach – from descriptive analytics (or the great human tradition of story telling) to the use of data to drive consideration of the future. These five stages have tremendous implications for how IoT is applied to business problems. The data can be used to consider the past (what happened, when did it happen, who was involved etc.) so it becomes a passive collection of historical record. In use cases like public safety and contracts, that timeline of definitive information is essential to understanding the road toward an outcome. The third and fourth stages of predictive and prescriptive analytics (there’s an important distinction between these two) take the data to develop insights as to the future outcomes and, more critically, to propose actions in response to those insights.

An example, perhaps. The historical weather record can be used to generate predictive weather outcomes based on the prevailing conditions, what it was like this day a week, month, year, century ago. That prediction becomes more accurate the closer a future time gets. As a result, it is easier (relatively!) to predict this afternoon’s weather than the weather a month from now.

Prescriptive analytics takes that prediction and then, based on what people or things did in response to it in the past, offers suggestions about what to do next. In this way, the analytics moves from a static partner in decision making (“here’s what’s going to happen”) to an active participant in driving the outcome (“here’s what might be a good idea”).

Once again, an example – historically, the current weather conditions suggest that a hurricane is due to make landfall in the next 12-24 hours. Predictive analytics offers the suggestion that it might make landfall, the speed and intensity of wind and the possibility of damage based on previous hurricanes in a similar environment. Prescriptive analytics then might cancel all leave for emergency services personnel, move non-critical patients from hospitals in the area and suggest hardware stores get in a supply of nails and plywood panels while also cancelling school and issuing actionable warnings. These actions are derived from activities which have been seen to be the most successful in minimizing loss of property and life in prior hurricanes, not only in the current vicinity but also in a broader area of hurricane affected locales.

As an organization moves from descriptive analytics (“On this day in 1924, a hurricane came through and flattened everything”) to predictive (“Based on all the data, there is a 90% likelihood of a hurricane hitting within 24 hours”) to prescriptive (“Batten down the hatches, hurricane a-coming”) the analytical model moves from being reference material to being partner in action.

And thus to “knowledge is power”.

Another example:

A global logistics and transportation company has implemented a vehicle fleet analytics solution by leveraging data generated through sensors and telematics. Previously, the vehicle maintenance schedule was based on a combination of usage and time-based considerations. This did not consider driving patterns, environmental factors, and onboard vehicle factors, including crank voltage, ignition voltage, and acceleration, all of which have a significant influence on the overall health of the engine and vehicle.

The organization observed that a large percentage of its vehicles broke down in transit because of electrical failure, with battery failure being a common cause. Battery failures result in unmet service-level agreements and costly readjustment of schedules to provide replacement vehicles. In this project scope, the enterprise collated and analyzed the historical maintenance records of 10,000 vehicles across three years of sensor data and fed it into a vehicle fleet analytics solution.

Based on its analysis, the company identified several major root causes and reasons for battery failures. It defined and executed a machine learning classifier that would run this data to predict an upcoming battery failure based on historical data trends and in-vehicle sensors.

This enabled the organization to track its fleet health proactively while reducing downtime and improving resource utilization because of avoidable service failures.

In climbing the ladder towards greater levels of analytical maturity (and the data required to fuel that maturity), it also becomes critical to transform the culture of the organization to become analytics consumers – unapplied knowledge which does not generate action is the data equivalent of potential versus kinetic energy.

One has the opportunity to change the world, the other just gets on and does it. Which one will you choose? 

For more use cases involving the collection and application of data to predictive outcomes, check out “IDC PeerScape: Practices for Connected Vehicles” by Shaily Shah and Hugh Ujhazy. For more from Dr. Kirk D Borne, please refer to

Hugh Ujhazy, associate vice president of IOT & Telecoms at IDCWritten by Hugh Ujhazy, associate vice president of IOT & Telecoms at IDC

Be the first to comment

What do you think?

This site uses Akismet to reduce spam. Learn how your comment data is processed.