This is the third article in a series focusing on a technology that is rising in importance to enterprise use of big data – IoT Analytics, or the analytical component of the Internet-of-Things. In this segment, we’ll provide an overview of the rise of IoT analytics. Previous parts to this special feature:
The Rise of IoT Analytics
IoT Analytics implies data, fast data, and big data. IoT is not just about capturing sensor data, or GPS locations, or temperature, or velocity changes. You have to find meaning in that data through analytics. There is much potential in terms of analytics that can be done on IoT data and there’s been an over obsession with collecting the data and managing the devices – but you also need the analytics to figure out the meaning in that data – to turn that into actionable insights for competitive advantage.
IoT analytics depends on the streaming analytics technology stack to deliver in-the-moment insights to differentiate the data. Predictive models know what to do in real-time, from a unified, flexible, analytics platform. Enterprise solutions that create predictive models via machine learning will be most valuable because the machine learning customizes the outcome for that enterprise based on its data.
Because IoT provides immediate data collection, businesses in all industries will benefit from improved decision making. IoT data sources generate vast amounts of fast-moving data. Successfully handling this so-called “fast data” is a serious challenge. With streaming analytics tools like Spark Streaming, being able to analyze and distribute intelligence faster means that tedious data collection will be a thing of the past. Decisions can be made faster, and in some cases can be automated. What this means in essence; is better decisions based on better data.
People routinely speak about the real-time aspect of IoT analytics, but it is really about “business-time.” It’s useful to divide things up into “right timing” rather than “real timing” of the business, for example – a customer walks into a shopping mall, electricity usage spikes at a factory, a shopper clicks on an online ad, a temperature sensor spikes, a stock price rises, a customer uses a credit card.
There are four types of analytics needed for IoT – and the execution platform that can actually use those analytics in real-time within IoT applications. It is important to deploy an analytics solution capable of supporting all four approaches:
- Descriptive analytics – starts with traditional BI where you gather all the data, expose it with a dashboard and reporting, and explore it with a data visualization tool. You want to see what’s happening. You want to cultivate human insights into the information being collected. This is where you’ll develop hypotheses about other applications and create more value in those applications.
- Predictive analytics – contains method for how can you find meaning in this data, and how you can you find a predictive model that foresees when a piece of equipment is going to fail for example. Here, you need statistical learning methods against that model.
- Streaming analytics – this is where you’re looking for patterns and looking for analytics that can be acted upon. But you don’t know what to look for until you do the descriptive analytics, unless you first have some human insight into what you’re looking for; and you can’t automate this process unless you have a predictive model. The result of the descriptive and the predictive phase finds its way into the streaming phase which you want to detect in real-time.
- Prescriptive analytics – at this stage you’ve detected the pattern or the event, now what do you want to do about it and when do you want to do it? This includes a whole other analytics regime like constraint based optimization, another predictive model, or customary business rules describing how to take that action.
Given the distributed nature of connected devices and the rapid growth of IoT infrastructures, we’re starting to see more organizations looking to execute analytics at the “edge.” As a result, the ability to push analytic capabilities to the source of data and run them directly will become paramount. Analytics at the edge, or performing analytics at the source versus within a central data warehouse, is receiving much attention in the industry by players in the IoT ecosystem working hard to make this clear competitive advantage a reality.
The opportunities are there for businesses who adopt IoT analytics today. As the volume of IoT data grows, so does the potential for insights. The benefits exist whether they seek to improve manufacturing efficiency, streamline logistics processes, or even provide new ways for customers to interact and receive information. So with respect to IoT coupled with analytics, the question is not why should we care, but is rather, can you afford not to? It’s all a matter of providing “digital intelligence” to the enterprise.
Contributed by Daniel D. Gutierrez, Managing Editor of insideBIGDATA. In addition to being a tech journalist, Daniel also is a practicing data scientist, author, educator and sits on a number of advisory boards for various start-up companies.
Sign up for the free insideBIGDATA newsletter.