Plugged in but disconnected
Bill Schmarzo, chief technology officer, EMC Global Services Big Data Practice, explains why being ‘connected’ doesn’t make us ‘smart’.
With the increasing hype around the Internet of Things (IoT) and the Internet of Everything (IoE), the potential enterprise opportunities are getting lost in endless discussions about the technologies that underpin the IoT and IoE (sensors, beacons, telematics, fitness trackers, mobile apps, global positioning devices, etc.). So, the question persists – how do we actually realize business value by tapping into IoT and IoE technologies?
As distinctly defined, Internet of Things is the maturation of the Internet in which everyday objects or devices (things) have network connectivity, allowing these objects or devices to send and receive data about their operations. Internet of Everything is a broader term that refers to devices and consumer wearable products connected to the Internet and outfitted with expanded digital features. Internet of Everything connects devices and humans.
Millions of different ‘connected’ devices and massive proliferation of data has culminated into lots of confusion. Unless, an organisation has its digital agenda underway to delineate the objective of processing and leveraging the wealth of data.
Being able to capture, store and manage the data created by connected devices and humans creates amazing possibilities. I, like many people, wear at least one fitness band, which tracks my steps, workout time, sleep, heart rate, and more. But that data by itself isn’t useful unless I apply some analytics to understand what it’s telling me about my health. I want to see trends, outliers, patterns and any other imbalances to help extract the right insights and recommendations to make better decisions about my diet, exercise, sleep, and overall health.
Let’s extend this example to the “connected” city, comprised of a wide range of devices (traffic lights, parking meters, weather instruments, etc.) and video cameras (traffic, pedestrian and bike traffic flow) generating data about city operations. A citizen could combine these sensors and video-generated data with other data sources, such as social media (Facebook, Instagram, Yelp) + citizen comments (emails, phone calls) + city reports (police blotters, fire reports, emergency services, construction permits, work orders, building hours, etc.) + local events (concerts, sporting events, farmers markets, parades, festivals, etc.) to create a rich perspective on the city’s activities, problems and overall economic and social vitality.
However, having a “connected” city does not mean that you have established a “smart” city. So how do we get smart?
Getting smart starts by understanding the city’s key business initiative or business objective (i.e., “what” we want to accomplish). For example, let’s identify and understand the decisions that city management (our key business stakeholder in this example) needs to make to support the business initiative of “improving traffic flow.” This could include: making traffic flow decisions, road repair and maintenance decisions, construction permits decisions, events management decisions etc. Each group of decisions equates into a use case, or the “how” we will accomplish the “what” of the business initiative.
The next step is to brainstorm the questions stakeholders need to answer in support of key decisions. This process will help to identify variables and metrics that might be better predictors of the decisions we are trying to make. While most organisations have a good handle on the ‘descriptive’ (what happened?) questions, the business stakeholders struggle with the “predictive” (what is likely to happen?) and the “prescriptive” (what should I do?) questions.
Brainstorming predictive and prescriptive questions typically uncovers numerous new data sources that are worthy of consideration. And this is a key point: all data sources are worthy of consideration!
Next, we assess the business value and implementation feasibility of each of the brainstormed data sources. This is where we determine the business value and the implementation feasibility, over the next 9 to 12 months, of each of the data sources vis-à-vis the use cases.
The final step is testing different analytic models that might yield the desired results that support effective decision-making. Data enrichment techniques such as RFM (Recency of activities, Frequency of activities, Monetary value of activities) will be employed to transform base metrics into potentially actionable metrics. It’s not unusual to test 10 to 20 different analytic models using the wealth of base and transformed metrics to isolate the ones that yield the best results.
For example, we might test Cluster Analysis to identify groups of drivers and/or events that impact traffic flow. Time Decomposition analytic algorithm to identify events that are driving traffic jams. Implement Behavioral Analytics to identify and quantify the impact in changes in drivers and traffic behaviors. Sentiment Analysis to analyses social media data to uncover areas of constituent dissatisfaction and under-performance.
Transitioning from “connected” to “smart” takes a lot of upfront work, but the more work that is invested in identifying, understanding and supporting the key decisions necessary to support the targeted business initiative, the more productive the data science will be.
In the end, becoming smarter is all about extracting value from the data generated through connected devices and by leveraging it for effective decision making.
Bill Schmarzo, chief technology officer, EMC Global Services Big Data Practice.