The functions and advantages described above are common knowledge to most organizations by now. But most organizations also fail to understand that being connected is one thing, being smart is something else entirely! All of these connected objects generate data. Data which, when analyzed, are a source of valuable information. I’d like to call this IoTtA: the Internet of Things to Analyze. The insights provided by the analysis of these data can be huge and lead to huge improvements in efficiency or productivity.
And, to take matters even one step further: storing the data that connected objects gather and analyzing them at a later point is just one of way of analyzing the Internet of Things. However, to truly harness the potential of the Internet of Things, we must be able to analyze the data as they are streaming to our a nalytics engines, in real-time. Managing, monitoring and analyzing live streaming data will open up endless possibilities for enterprises. If anyone still needs convincing, the examples below may provide some inspiration.
A technology where streaming analytics is about to break through is image recognition. We will see some pilot projects starting at the end of this year. As soon as they have proved their value, things may move fast.
Applications for image recognition are not hard to come up with. Governments, for instance, can use image recognition for security purposes, such as monitoring passengers in an airport terminal and identifying suspicious individuals in real time.
Image recognition can also be used by retailers who are striving for a customer-oriented approach in their stores. Sensors can be used to monitor everyone who walks into the store. When streaming analytics are in place, faithful customers can be identified as they walk through the door. This will make it easier than ever before to greet a customer in person the moment he steps inside the store, and to make them a customized offer they can’t refuse!
And the future does not end here. Some early adopters are developing image recognition applications to recognize and identify the spare parts (including details about the specific type and model) of, say, a truck or a washing machine. This will enable users to simply take a picture of a defective spare part and let the app come up with all relevant identification information. This information can then be sent to a warehouse, where the correct spare part can be selected in the blink of an eye and sent to the user.
A related technology where streaming analytics can be useful involves voice analysis.
SAS has developed a proof of concept for a call centre that was responsible for managing potentially problematic loans. The call centre had a standardized process in place that left lots of room for improvement in terms of effectiveness and speed: calling the debtor, sending a letter, sending a registered letter, and so forth, until, finally, a bailiff order was issued to collect unpaid debts.
The proof of concept showed that using voice analysis could sometimes shorten the procedure and lead to a higher succes rate. If the analysis of the tone and vocabulary of the initial phone call (the first step in the standardized procedure) suggested that the normal procedure might not be efficient, the call center could decide to bypass certain steps which, for similar calls in the past, proved useless and ineffective. Avoiding some of the intermediate steps and shortening the time span of the standardized procedure benefits both the debtor and the creditor.
Connected Truck at SAS Forum
Another example of the possibilities offered by streaming analytics was provided at the SAS Forum in Amsterdam earlier this fall. We presented a miniature prototype of a connected truck, which we developed together with Intel and Mendix. If asked what the purpose of the truck is, the answer is simple: to predict the future!
The truck is equipped with sensors, and with a server and SAS software which can analyze the data that stream in through the sensors in real time. Anything that seems relevant can be sent to the truck owner/manufacturer. The truck has an integrated app which allows the company to view the location of the truck as well as various statistics, ranging from oil temperature to the speed at which the truck is driving. It will also send alerts whenever something is going wrong. So if the sensors measure anything suspicious, the truck company can be warned even before any defect has occurred.
A similar technology would also be useful for wind turbines. Although, so far, they have been spared from serious defects, a wind turbine is a complex object that can get damaged or defective in many ways, with serious consequences. Sensors and streaming analytics would allow us to predict failures before they take place and take the appropriate actions.
Or how about hospitals, which could use sensors to monitor patients in their cherished and familiar home environment, rather than keeping them in the hospital?
In short: anything that can be followed live is a potential source of streaming data!
Streaming Analytics: There’s More to It Than You Think
Streaming analytics is very different from regular data analytics, and requires an analytical maturity which few companies have reached. Most organizations still struggle with basic questions such as: how long should we wait before analyzing the data, and what kind of analysis should we conduct?
The road ahead may indeed be long, but you can rest assured that streaming analytics - of IoT objects as well as from ‘regular’ sources - will be a key to success in the future.