Subscribe to DSC Newsletter

The Case for a New “Final Frontier” in Data Analytics

This article has been contributed by Alain Louchez (Georgia Tech Research Institute)

The Internet of Things already integrates a new phase beyond prescriptive analytics. 

There is no shortage of attention lately on the “Internet of Things”. As a case in point, see the “Developing Innovation and Growing the Internet of Things Act” or “DIGIT Act”, i.e., S. 2607, a  bill introduced in the Senate on March 1, 2016 and amended on September 28, 2016,  “to ensure appropriate spectrum planning and inter-agency coordination to support the Internet of Things” – A companion bill, H.R. 5117, was introduced in the House of Representatives on April 28, 2016.

However, since there is no “internet” dedicated to “things”, it is fair to state that the Internet of Things does not exist as such.  We are left with a definitional vacuum, but it is hammering the obvious to acknowledge that there is no dearth of attempts around the world to fill the gap. Perhaps as a helpful shortcut, we could view the expression as a metaphor that captures the arrival of almost anything and everything, until now out of scope, into the communications space.

Source for picture: click here

While the proposed DIGIT Act sees the Internet of Things as referring to the “growing number of connected and interconnected devices”, we could argue that the smartness of those devices is what makes IoT truly unique; hence the “interconnection of intelligent things” may be a more accurate descriptor.

In sum, the term “Internet of Things” heralds the advent of what we could call a true “pulsating world” arising from sending data from and to smart devices.

Extracting information, and therefore value, from the data captured at the edge of the Internet of Things network is increasingly becoming a core focus of IoT solution providers.

As the Internet of Things is bound to become a gargantuan reservoir of data, it close interaction with data analytics has been well documented in academia and business.

Charles McLellan writing for a special feature of ZDNet in March 2015 on the Power of IoT and Big Data: Unlocking the Power, agrees that “the IoT will massively increase the amount of data available for analysis by all manner of organizations,” but cautions that “the volume, velocity and variety (not to mention variable veracity) of IoT-generated data makes it no easy task to select or build an analytics solution that can generate useful business insight.”

Business Analytics frameworks such as IBM’s and Gartner’s are widely used to chart the course of the business insight extraction. The former identifies three analytic capabilities, i.e., descriptive, predictive and prescriptive while the latter inserts a diagnostic stage between descriptive and predictive. Table 1 summarizes both frameworks:

Table 1: Existing Business Analytics Categories

Business Analytics Category



Descriptive Analytics

"What has happened?"

"What happened?

Diagnostic Analytics


“Why did it happen?

Predictive Analytics

"What could happen?”

"What will happen?”


"What should we do?”

“How can we make it happen?


Extending the data analytics discussion to the Internet of Things (a.k.a. “the analytics of things”), Professor Thomas Davenport, in a March 2016 Data Informed article, wonders When Will the Analytics of Things Grow Up? He expresses the concern that “most of the ‘analytics of things’ thus far have been descriptive analytics – bar (and Heaven forbid, pie) charts, means and medians, and alerts for out-of-bounds data,” and highlights areas where business analytics can make a difference in IoT beyond descriptive (dashboard-type report on performance) such as diagnostic (alerts that need attention), predictive (e.g., breakdown potential) and prescriptive (recommendations based on predictions, experiments, or optimizations).

Prescriptive has often been described as the “final frontier” of data analytics. Yet, or perhaps because of this, a recent academic paper observes that “there are very limited examples of good prescriptive analytics in the real world” (1). The same paper proposes an overall definition “in general, prescriptive solutions assist business analysts in decision-making by determining actions and assessing their impact regarding business objectives, requirements, and constraints. For example, what if simulators have helped provide insights regarding the plausible options that a business could choose to implement in order to maintain or strengthen its current position in the market.” Let’s note in passing the “assisting” role of said solutions. Indeed advising outcomes and providing recommendations are presently associated with this final phase of business analytics. This seems to somewhat contradict the notion of “prescription” since its very meaning implies imposition of one single direction or rule. However, the current generally-accepted use gives room for interpretation or non-compliance.

Professor Davenport’s conclusion in the above-mentioned article anticipates the need for another type of prescription in the IoT world: “In some IoT environments such as smart cities analytics will need to provide automated prescriptive action /…/ In such settings, the amount of data and the need for rapid decision making will swamp human abilities to make decisions on it.”

Davenport’s prescient clarity is strengthened when we realize that, in the IoT space, the future, with “automated prescriptive action”, is actually already here (e.g., IoT devices with actuators, industrial robots, etc.). Consequently, we might have to recognize a possible fifth phase (a new final frontier?), i.e., normative analytics (or automated prescription), whose conceptual tenets lean on system engineering. For all intents and purposes, normative analytics is a self-adaptive system, which adjusts dynamically to changes in external and internal conditions (a point of comparison could be rocket launch technologies).

While not yet mainstream, the following two examples with life and death consequences, i.e., driverless cars and autonomous robots for surgery, show the interactive analytics continuum.

Data analysis for driverless cars (or more generally, unmanned vehicles) has to go through at least five stages, i.e., 1) descriptive (e.g., wear and tear of a multitude of components); 2) diagnostic (e.g., what’s wrong?); 3) predictive (e.g., impact of traffic jams, weather, detours); 4) prescriptive (based on previous analytics layers, what acceptable options does the vehicle have?) and finally 5) a normative stage (what the vehicle must do). All these phases must happen concurrently and seamlessly; there is no time to delay the final decision, i.e., the car must stop or move. The consequences of a delay, however minute, could be catastrophic.

The same can be said of surgical robots. For instance, a team of researchers at the Sheikh Zayed Institute for Pediatric Innovation within the Children’s National Health System, and Johns Hopkins University recently announced that they had developed a “Smart Tissue Autonomous Robot” (STAR) for soft tissue surgery, which is “a difficult task for a robot given tissue deformity and mobility” (2). Quite remarkably, they claim that they have demonstrated that “supervised autonomy with STAR not only is feasible but also, by some metrics, surpasses the performance of accepted surgical procedures including RAS (Robot-Assisted Surgery), LAP (Laparoscopy), and manual surgery” and that “autonomous surgery can bring better efficacy, safety, and access to the best surgical techniques regardless of human factors, including surgeon experience.” Is there a better example of automated and integrated analytics intimately connected with action?

The above two IoT-related use cases exemplify the smooth transition from the virtual to the physical world through analytics and may be used as a template/model for business analytics (see Table 2).

Table 2: Analytics Categories with a New “Final Frontier”

Business Analytics Category


Category          Differentiator

Human Factor

Descriptive Analytics

"What has happened"    (Inform)

"It is what it is"

Except for the selection of statistical techniques and areas of interest, does not shape the output

Diagnostic Analytics

“Why did it happen?” (Understand)

“This is what needs attention”

Except for the selection of statistical techniques and areas of interest, does not shape the output

Predictive Analytics

"What could happen?"    (Forecast)

"This is what the future may look like"

Except for the selection of statistical techniques and areas of interest, does not shape the output

Prescriptive Analytics

"What should we do?"       (Advise)

"These are the optimal options we have" - Possible delay between analysis and decision

Shape the output, but post-analysis - Human decides on execution of potential actions.

Normative Analytics

"This is what must happen."                 (Execute)

Based on predictive analytics, next step is automatically triggered and instrumented - No delay between analysis and decision/action

Shape the output with action criteria embedded in this phase - Except for monitoring, human is not involved in execution.




Occam’s razor notwithstanding, there is a case to be made for specifying a new phase in data analytics, i.e., normative analytics, which, whether identified or not, already exists, especially in the Internet of Things universe. This is the step where, fed by the other analytics phases, action is automatically triggered and executed.

It would be hard to merge prescriptive analytics (as currently defined in the analytics space) with the proposed normative analytics. There will always be situations where the need for injecting human input into the final decision preempts process automation. In any event, prescriptive analytics remains the necessary step that must be gone through and successfully passed before normative analytics kicks in.

Note that this model can be applied to any Cyber-Physical System (CPS), whatever its size. The National Institute of Standards and Technology (NIST) defines Cyber-Physical Systems as “smart systems that include engineered interacting networks of physical and computational components,” and underscores that “CPS and related systems (including the Internet of Things (IoT) and the Industrial Internet) are widely recognized as having great potential to enable innovative applications and impact multiple economic sectors in the worldwide economy”. It also foresees that CPS’ new capabilities will fuse with other important evolutions of technologies such as big data analytics, which “are expected to bring transformational changes to economies, societies, our knowledge of the world, and ultimately the way people live” (3).

Dr. Frey and Professor Osborne’s abundantly-quoted study on computerization serves as a reminder that the broad diffusion of automated prescription might just be around the corner. According to their estimate, “47 percent of total US employment is in the high risk category, meaning that associated occupations are potentially automatable over some unspecified number of years, perhaps a decade or two” (4).

This prospect cannot materialize without normative analytics resting on optimization techniques. The irony is that as high technology increasingly pervades all economic functions, non-technology considerations become part and parcel of the whole framework. Policy, regulation, ethical and other critical perspectives need to be factored in the optimization calculation.

Among its many applications, normative analytics could open the door to a new regulatory domain, which has been called “governance by things” (5) or “algorithmic regulation” (6), i.e., a new way of implementing rules and norms through analytics not, however, without misgivings (7).

Regardless of the attractiveness of this new concept, the introduction of normative analytics as an integral phase of the data analytics spectrum must be further validated, and prudence advocated the same way William Vorhies not too long ago cautioned against the possible groundless addition of prescriptive analytics in the data analytics nomenclature, i.e., “with about 1/3rd of companies having yet adopted predictive analytics the last thing we need is introducing ‘the next big thing’ into that conversation unless there is some real value to be had” (8).


(1)    Uthayasankar Sivarajah, Muhammad Mustafa Kamal, Zahir Irani, and Vishanth Weerakkody, “Critical analysis of Big Data challenges and analytical methods”, Journal of Business Research, Elsevier, August 10, 2016, available at


(2)    Azad Shademan, Ryan S. Decker, Justin D. Opfermann, Simon Leonard, Axel Krieger, and Peter C. W. Kim, May 4, 2016, “Supervised autonomous robotic soft tissue surgery”, Science Translational Medicine 8 (337), available at


(3)    See Framework for Cyber-Physical Systems – Release 1.0 – May 2016, document prepared by the Cyber-Physical Systems Public Working Group (CPS PWG), an open public forum established by the National Institute of Standards and Technology (NIST), available at


(4)    Carl Benedikt Frey, and Michael A. Osborne, “The Future of Employment: How Susceptible are Jobs to Computerization?”, September 17, 2013, Oxford Martin Press, available at


(5)    Wolfgang Shultz and Kevin Dankert, ’Governance by Things’ as a challenge to regulation by law, Internet Policy Review, Volume 5, Issue 2, June 30, 2016, available at


(6)    Brett Goldstein, “When Government Joins the Internet of Things”, The New York Times, September 8, 2013, available at


(7)    Evgeny Morozov, “The Rise of Data and the Death of Politics, The Guardian (UK), July 19, 2014, available at


(8)    William Vorhies, “Prescriptive versus Predictive Analytics - A Distinction without a Difference?”, October 23, 2014, Decision Science Central, available at



The views expressed in this article are solely the author’s and do not necessarily represent those of the Georgia Institute of Technology (“Georgia Tech”), the Georgia Tech CDAIT members, the University System of (U.S. State of) Georgia or the (U.S.) State of Georgia.

Alain Louchez is the Managing Director of the Center for the Development and Application of Internet of Things Technologies (CDAIT pronounced “sedate”) at the Georgia Institute of Technology (“Georgia Tech”), Atlanta, Georgia. CDAIT’s purpose is to expand and promote the Internet of Things (IoT)’s huge potential and transformational capabilities through research, education and industry outreach. The Center is sponsored by global companies headquartered in North America, Europe, Asia and Australia.  Alain was recently selected by Instituto Tecnológico y de Estudios Superiores de Monterrey (Guadalajara) as an international advisor to “Centro de Innovación, Desarrollo Tecnológico y Aplicaciones de Internet de las Cosas” (a.k.a. “Center of Innovación in Internet of Things or CIIoT”) after bid approval by the government of Mexico in June 2016. Prior to joining Georgia Tech, Alain held various executive positions including member of the board of directors of leading companies in the high tech industry, in Europe and the United States.

Views: 1436


You need to be a member of Data Science Central to add comments!

Join Data Science Central

Follow Us


  • Add Videos
  • View All


© 2017   Data Science Central   Powered by

Badges  |  Report an Issue  |  Privacy Policy  |  Terms of Service